Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
DETAILED ACTION
Status of the Application
The following is a Final Office Action.
In response to Examiner's communication of 8/5/2025, Applicant responded on 10/31/2025. Amended claim 21-23, 28, 29, 32, 34, 37, and 39.
Claims 21-40 are pending in this application and have been examined.
Response to Amendment
Applicant's amendments to claims 21-23, 28, 29, 32, 34, 37, and 39 are sufficient to overcome the 35 USC 112 rejections set forth in the previous action. However, Applicant’s amendments necessitated new grounds of rejections.
Applicant's amendments to claims 21-23, 28, 29, 32, 34, 37, and 39 are not sufficient to overcome the 35 USC 101 rejections set forth in the previous action.
Applicant's amendments to claims 21-23, 28, 29, 32, 34, 37, and 39 are not sufficient to overcome the prior art rejections set forth in the previous action.
Response to Arguments – 35 USC § 101
Applicant’s arguments with respect to the rejections have been fully considered, but they are not persuasive.
Applicant submits, “…Applicant's amended claims set forth a specific well-defined computer-implemented process to generate and train neural networks for runtime deployment that identify significant events in an environment. The claimed process systematically processes digital data to train the neural network and automatically update the encoded strength of certain gates within the network, thereby enhancing its real-time accuracy during runtime…These features collectively define a concrete and technologically-rooted process that improves the functioning of the neural network in a specific application domain. The claims do not merely recite an abstract idea or mathematical concept, but instead describe a practical implementation with clearly defined structural and functional limitations. Here, the claimed training phase along with runtime execution of the resulting neural network cooperate to produce a technological result that is far more than the sum of individual steps. An analysis by the Examiner that excises these inter-dependent limitations would counter the guidance and requirements provided by the USPTO Memo on Reminders…the published USPTO examples 39, which illustrates claim limitations that merely involve an abstract idea, and 47, which shows limitations that recite an abstract idea. The claim limitation "training the neural network in a first stage using the first training set" of example 39 does not recite a judicial exception. Even though "training the neural network" involves a broad array of techniques and/or activities that may involve or rely upon mathematical concepts, the limitation does not set forth or describe any mathematical relationships, calculations, formulas, or equations using words or mathematical symbols... the Applicant's amended independent claims 21, 32, and 39 recite a distinctive computer-implemented process for training and updating a neural network for runtime execution in a specific application domain. These claims describe a structured, iterative approach to improving the neural network performance by adjusting the encoded strength of gates based on labeled training data and probabilistic comparisons...As illustrated in USPTO Example 39, a claim limitation such as "training the neural network in a first stage using the first training set" was found to merely involve an abstract idea without reciting one, precisely because it did not reference any specific mathematical operations. Applicant's claims follow this same pattern: they describe the functional behavior and structural configuration of a neural network in a way that is technologically meaningful, without invoking named algorithms or mathematical constructs that would trigger further eligibility scrutiny under Step 2A. Accordingly, Applicant's pending claims are analogous to those in Example 39 and should be considered patent eligible under the USPTO's current guidance. They involve, but do not recite, judicial exceptions, and therefore do not require further eligibility analysis… Applicant's pending claims cannot be practically performed in the human mind. As clarified in the USPTO Memo on Reminders, "a claim does not recite a mental process when it contains limitation(s) that cannot practically be performed in the human mind, for instance when the human mind is not equipped to perform the claim limitation(s)." USPTO Memo on Reminders at p.2. While the human mind may be capable of recognizing a weather condition as severe based on experience or observation, this is not what the Applicant's claims recite. Instead, the pending claims describe a specific, computer-implemented data-driven process involving the generation and training of a neural network using labeled data, probabilistic modeling, and dynamic gate-strength adjustments. These are all operation that require computational resources and iterative data processing far beyond the capabilities of the human mind. Applicant's disclosure teaches a combination of unique data-driven techniques that are inherently computational in nature, implemented by a computing system to automatically and accurately identify significant events, such as weather anomalies, in an environment. These operations involve processing large volumes of data across multiple dimensions (e.g., time, location, event type), generating probability factors, and updating internal network parameters, all of which are tasks that cannot be performed mentally, either practically or reliably...When viewed holistically, Applicant's claims describe a technical solution to a real-world problem using a machine-learning architecture that operates in a manner fundamentally distinct from human cognition. For at least these reasons, the Applicant' s claims as a whole do not recite a mental process and therefore are patent eligible under current USPTO guidance…Applicant respectfully submits that the currently pending claims are directed to a specific technological improvement in both neural network training and event detection systems (e.g., weather event detection systems) and thus are patent eligible. Applicant's pending claims, reflected above, improve two distinct technological areas: (i) neural network construction, through a unique process of updating encoded strengths of gates within a neural network, which enhances training efficiency and accuracy beyond conventional approaches, and (ii) event detection systems, by enabling real-time automated detection of significant environmental events, such as severe weather, before they escalate into conditions that may cause irreparable damage or harm to individuals. These concrete improvements satisfy the "practical application" inquiry mandated by the USPTO Memo on Reminders, as they are rooted in specific technological advancements rather than abstract ideas. More specifically, Applicant's claims are directed to a data-driven, automated system for detecting significant events using a uniquely trained and dynamically updated neural network…Conventional techniques for event detection often rely heavily on human input, manual rule- based systems, or static thresholding methods that lack adaptability. These approaches are limited in their ability to respond to rapidly changing or complex environments, resulting in inefficiencies and reduced accuracy. Applicant's claimed invention addresses these shortcomings by providing processes that continuously learn from labeled data and adjusts internal parameters to improve detection precision. This eliminates reliance on human judgment and reduces risk of oversight or delay in identifying critical events. The claimed technology thus offers a proactive and scalable solution that enhances both the reliability and responsiveness of event detection systems…which often suffer from limitations such as overfitting, slow convergence, and poor generalization to unseen data. Traditional methods may sacrifice accuracy in favor of computational simplicity or rely on fixed training algorithms that do not adapt to the evolving nature of input data. These methods are also prone to errors such as misclassification, instability during training, and inability to handle noisy or incomplete datasets. By contrast, Applicant's approach introduces a dynamic gate-strength updating mechanism that allows the neural network to refine its internal structure based on probabilistic comparisons with labeled data. This results in more accurate and robust performance in real-world scenarios, particularly in environments where conditions change rapidly and unpredictably. For at least these reasons, and as further supported by the USPTO Memo on Reminders, Applicant's pending claims integrate any alleged judicial exceptions into a practical application and therefore are patent eligible under Step 2A Prong Two and current USPTO guidance…in Bascom, the combination of features in Applicant's amended claims is non- conventional and non-routine. None of Applicant's claimed features are conventional or routine in the field of neural network-based event detection systems, particularly those designed for real-time environmental monitoring. Moreover, the combination of these claimed features, which includes training a neural network using labeled data, generating probabilistic assessments of event significance, and dynamically updating gate strengths within the network to improve runtime accuracy, is neither routine nor conventional. This integrated approach departs from traditional machine learning pipelines, which typically rely on static architectures and fixed training algorithms. The result of this combination is an improvement to both the underlying neural network architecture and the accuracy and responsiveness of event detection systems deployed in dynamic environments. By enabling real-time adaptation and reducing reliance on human oversight, Applicant's invention provides a technical solution to a longstanding problem in the field. Even assuming arguendo that each claimed element was known, the ordered combination of features recited in Applicant's independent claims is precisely the type of non- conventional arrangement that the Court and the USPTO Memo on Reminders identify as supplying "significantly more."….The USPTO Memo on Reminders explains:" Examiners are reminded that if it is a 'close call' as to whether a claim is eligible, they should only make a rejection when it is more likely than not (i.e., more than 50%) that the claim is ineligible under 3 5 U.S.C. 101. A rejection of a claim should not be made simply because an examiner is uncertain as to the claim's eligibility. In order to make a rejection of a claim under any of the statutory bases...unpatentability must be established by a preponderance of the evidence." See USPTO Memo on Reminders at p.5 (emphasis added). Applicant respectfully submits that the Examiner has not met the burden of establishing unpatentability by a preponderance of the evidence. When the currently pending claims are properly considered as a whole, including their specific combination of technical features and the improvements they provide to neural network training and event detection systems, it is more likely than not that the claims are patent eligible...” The Examiner respectfully disagrees.
Although, Applicant’s amendments furthers prosecution, unlike Example 39 and BASCOM, by Applicant’s own admission in Applicant’s remarks, the claims are directed to, …weather prediction… automatically and accurately identify significant events, such as weather anomalies, in an environment…human mind may be capable of recognizing a weather condition as severe based on experience or observation…Conventional techniques for event detection often rely heavily on human input, manual rule- based systems, or static thresholding methods that lack adaptability...enabling real-time automated detection of significant environmental events, such as severe weather, before they escalate into conditions that may cause irreparable damage or harm to individuals…, which is a problem directed to a mental process (i.e. human observing weather, human predicting weather events based on human observation) and mathematical concepts (i.e. human calculating and predicting weather events by training mathematical models and using probabilities to predict significant weather events), as established in Step 2A Prong 1. This problem does not specifically arise in the realm of computer technology, but rather, this problem existed and was addressed long before the advent of computers. Thus, the claims do not recite a technical improvement to a technical problem or necessarily roots in computing technologies. Pursuant to the broadest reasonable interpretation, as an ordered combination, each of the additional elements are computing elements recited at high level of generality implementing the abstract idea, and thus, are no more than applying the abstract idea with generic computer components. Further, these additional elements generally link the abstract idea to a technical environment, namely the environment of a computer and machine learning, performing extra solution activities. Therefore, as a whole, the additional elements do not integrate the abstract ideas into a practical application in Step 2A Prong 2 or amount to significantly more in Step 2B. Even novel and newly discovered judicial exceptions are still exceptions, despite their novelty. July 2015 Update, p. 3; see SAP America Inc. v. Investpic, LLC, No. 2017-2081, slip op. at 2 (Fed Cir. May 15, 2018).
Simply reciting specific limitations that narrow the abstract idea does not make an abstract idea non-abstract. 79 Fed. Reg. 74631; buySAFE Inc. v. Google, Inc., 765 F.3d 1350, 1355 (2014); see SAP America at p. 12. As discussed in SAP America, no matter how much of an advance the claims recite, when “the advance lies entirely in the realm of abstract ideas, with no plausibly alleged innovation in the non-abstract application realm,” “[a]n advance of that nature is ineligible for patenting.” Id. at p. 3.
Use of a computer or other machinery in its ordinary capacity for economic or other tasks (e.g., to receive, store, or transmit data) or simply adding a general purpose computer or computer components after the fact to an abstract idea (e.g., a fundamental economic practice or mathematical equation) does not integrate a judicial exception into a practical application or provide significantly more. See Affinity Labs v. DirecTV, 838 F.3d 1253, 1262, 120 USPQ2d 1201, 1207 (Fed. Cir. 2016) (cellular telephone); TLI Communications LLC v. AV Auto, LLC, 823 F.3d 607, 613, 118 USPQ2d 1744, 1748 (Fed. Cir. 2016) (computer server and telephone unit). Similarly, “claiming the improved speed or efficiency inherent with applying the abstract idea on a computer” does not integrate a judicial exception into a practical application or provide an inventive concept. Intellectual Ventures I LLC v. Capital One Bank (USA), 792 F.3d 1363, 1367, 115 USPQ2d 1636, 1639 (Fed. Cir. 2015).
Response to Arguments – Prior Art
Applicant’s arguments with respect to the rejections have been fully considered, but they are not persuasive. However, Applicant’s arguments are moot in light of new grounds of rejections necessitated by Applicant’s amendments.
Claim Objections
Claim 22 is/are objected due to the following informalities.
Claim 22 recite “…predetermined threshold values value for the weather events…”. Appropriate corrections required.
Claim Rejections - 35 USC § 112(b)
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claim 10 are rejected under is/are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as failing to set forth the subject matter which the inventor or a joint inventor, or for pre-AIA the applicant(s) regard as their invention.
Claim 10 recites “…for display via a graphical user interface.”, it is unclear if this element is referring to “…for reviews via a graphical user interface.” in claim 1. Appropriate correction required.
Claim Rejections – 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 21-40 is/are rejected under 35 U.S.C. 101 because the claimed invention is directed to non-statutory subject matter.
Claim 21 recite, A … method for automatically identifying significant events in an environment, the method comprising:
during a training phase:
training a … to automatically identify significant events in the environment, wherein the …, the training comprising:
receiving training data that comprises labeled weather event data for a plurality of locations over a plurality of time periods;
generating, by the …, a probability factor representing a chance that a weather event in the training data will be significant for at least one location amongst the plurality of locations; and
comparing the generated probability to the labeled weather event data of the received training data; and
in response to the comparison, updating the …;
during a runtime phase:
receiving data from a plurality of …, wherein each of the … is configured to generate weather event data about a respective one or more locations, the weather event data comprising environmental measurements for inferring significance of weather events at the one or more locations;
providing the weather event data as inputs to the trained …;
receiving, as output from the trained …, weather event data having a respective significance level that is greater than a respective predetermined threshold value; and
correlating weather events with the outputted weather event data having the respective significant level that is greater than the respective predetermined threshold value.
Claim 32 recite, A … method for automatically identifying significant events in an environment, the method comprising:
training a … model to automatically identify significant events in the environment, wherein the … model was trained using training data that was labeled with significant events over a plurality time periods for a plurality of locations;
receiving, from a plurality of …, events data for one or more locations, the events data comprising environmental measurements for inferring significance of events at the one or more locations;
providing the events data as inputs to the trained … model;
receiving, as output from the trained … model, indications of the events associated with the events data having a respective significance level that is greater than a respective predetermined threshold value; and
returning the indications of the events having the respective significant level that is greater than the respective predetermined threshold value.
Claim 39 recite, A … method for automatically identifying significant events in an environment, the method comprising:
during a training phase:
training a … to automatically identify significant events in the environment, wherein the …, the training comprising:
receiving training data that comprises labeled weather event data over a plurality of time periods;
generating, by the neural network, a probability factor representing a chance that an event in the training data will be significant over at least one of the plurality of time periods; and
comparing the generated probability to the labeled weather event data of the received training data; and
in response to the comparison, updating the …;
during a runtime phase:
receiving data from a plurality of …, wherein each of the … is configured to generate event data that comprises environmental measurements for inferring significance of real-world events;
providing the event data as inputs to the trained neural network;
receiving, as output from the trained …, event data having a respective significance level that is greater than a respective predetermined threshold value; and
correlating the real-world events with the outputted event data having the respective significant level that is greater than the respective predetermined threshold value.
Analyzing under Step 2A, Prong 1:
The limitations regarding, …automatically identifying significant events in an environment…during a training phase: training a … to automatically identify significant events in the environment, wherein the …, the training comprising: receiving training data that comprises labeled weather event data for a plurality of locations over a plurality of time periods; generating, by the …, a probability factor representing a chance that a weather event in the training data will be significant for at least one location amongst the plurality of locations; and comparing the generated probability to the labeled weather event data of the received training data; and in response to the comparison, updating the …; during a runtime phase: receiving data from a plurality of …, wherein each of the … is configured to generate weather event data about a respective one or more locations, the weather event data comprising environmental measurements for inferring significance of weather events at the one or more locations; providing the weather event data as inputs to the trained …; receiving, as output from the trained …, weather event data having a respective significance level that is greater than a respective predetermined threshold value; and correlating weather events with the outputted weather event data having the respective significant level that is greater than the respective predetermined threshold value… training a … model to automatically identify significant events in the environment, wherein the … model was trained using training data that was labeled with significant events over a plurality time periods for a plurality of locations; receiving, from a plurality of …, events data for one or more locations, the events data comprising environmental measurements for inferring significance of events at the one or more locations; providing the events data as inputs to the trained … model; receiving, as output from the trained … model, indications of the events associated with the events data having a respective significance level that is greater than a respective predetermined threshold value; and returning the indications of the events having the respective significant level that is greater than the respective predetermined threshold value… during a training phase: training a … to automatically identify significant events in the environment, wherein the …, the training comprising: receiving training data that comprises labeled weather event data over a plurality of time periods; generating, by the neural network, a probability factor representing a chance that an event in the training data will be significant over at least one of the plurality of time periods; and comparing the generated probability to the labeled weather event data of the received training data; and in response to the comparison, updating the …; during a runtime phase: receiving data from a plurality of …, wherein each of the … is configured to generate event data that comprises environmental measurements for inferring significance of real-world events; providing the event data as inputs to the trained neural network; receiving, as output from the trained …, event data having a respective significance level that is greater than a respective predetermined threshold value; and correlating the real-world events with the outputted event data having the respective significant level that is greater than the respective predetermined threshold value…, under the broadest reasonable interpretation, can include a human using their mind and using pen and paper to perform the identified limitations; therefore, the claims are directed to a mental process.
Further, …automatically identifying significant events in an environment…during a training phase: training a … to automatically identify significant events in the environment, wherein the …, the training comprising: receiving training data that comprises labeled weather event data for a plurality of locations over a plurality of time periods; generating, by the …, a probability factor representing a chance that a weather event in the training data will be significant for at least one location amongst the plurality of locations; and comparing the generated probability to the labeled weather event data of the received training data; and in response to the comparison, updating the …; during a runtime phase: receiving data from a plurality of …, wherein each of the … is configured to generate weather event data about a respective one or more locations, the weather event data comprising environmental measurements for inferring significance of weather events at the one or more locations; providing the weather event data as inputs to the trained …; receiving, as output from the trained …, weather event data having a respective significance level that is greater than a respective predetermined threshold value; and correlating weather events with the outputted weather event data having the respective significant level that is greater than the respective predetermined threshold value… training a … model to automatically identify significant events in the environment, wherein the … model was trained using training data that was labeled with significant events over a plurality time periods for a plurality of locations; receiving, from a plurality of …, events data for one or more locations, the events data comprising environmental measurements for inferring significance of events at the one or more locations; providing the events data as inputs to the trained … model; receiving, as output from the trained … model, indications of the events associated with the events data having a respective significance level that is greater than a respective predetermined threshold value; and returning the indications of the events having the respective significant level that is greater than the respective predetermined threshold value… during a training phase: training a … to automatically identify significant events in the environment, wherein the …, the training comprising: receiving training data that comprises labeled weather event data over a plurality of time periods; generating, by the neural network, a probability factor representing a chance that an event in the training data will be significant over at least one of the plurality of time periods; and comparing the generated probability to the labeled weather event data of the received training data; and in response to the comparison, updating the …; during a runtime phase: receiving data from a plurality of …, wherein each of the … is configured to generate event data that comprises environmental measurements for inferring significance of real-world events; providing the event data as inputs to the trained neural network; receiving, as output from the trained…, are mathematical concepts.
Accordingly, the claims are directed to a mental process, mathematical concepts, and thus, the claims are directed to an abstract idea under the first prong of Step 2A.
Analyzing under Step 2A, Prong 2:
This judicial exception is not integrated into a practical application under the second prong of Step 2A.
In particular, the claims recite the additional elements beyond the recited abstract idea identified under Step 2A, Prong 1, such as:
Claim 21, 32, 39: computer-based, training a neural network, neural network comprises a plurality of artificial cells interconnected via a plurality of gates, and wherein each gate encodes a strength of a relationship in the connection between an output of one artificial cell and an input of another artificial cell, updating the encoded strength of at least some of the plurality of gates of the neural network, sensors, machine learning
Claim 27: satellites
Claim 38: Long Short-Term Memory (LSTM) model
, and pursuant to the broadest reasonable interpretation, as an ordered combination, each of the additional elements are computing elements recited at high level of generality implementing the abstract idea, and thus, are no more than applying the abstract idea with generic computer components. Further, these additional elements generally link the abstract idea to a technical environment, namely the environment of a computer.
Additionally, with respect to, “…receiving…”, “…generating…”, “…providing…”, “…returning…”, “…publishing…”, these elements do not add a meaningful limitations to integrate the abstract idea into a practical application because they are extra-solution activity, pre and post solution activity - i.e. data gathering – “…receiving…”, “…providing…”, “…returning…”, data output – “…generating…”, “…returning…”, “…publishing…”
Analyzing under Step 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception under Step 2B.
As noted above, the aforementioned additional elements beyond the recited abstract idea are not sufficient to amount to significantly more than the recited abstract idea because, as an order combination, the additional elements are no more than mere instructions to implement the idea using generic computer components (i.e. apply it).
Additionally, as an order combination, the additional elements append the recited abstract idea to well-understood, routine, and conventional activities in the field as individually evinced by the applicant’s own disclosure, as required by the Berkheimer Memo, in at least:
[0021] Once the model is trained, it may be put into operation. In some embodiments, for each new event occurring on a particular field, the event analysis module may calculate a probability for whether the event has significance in terms of traceability. If the calculated probability "P" is greater than a predefined threshold (e.g., P > 0.8) , then the event may be automatically posted to an ledger. In some embodiments, the ledger may be a distributed ledger, such as a blockchain.
[0046] For illustrative purposes, the ML model 440 will be described with reference to a LSTM class ML model ("LSTM model"). LSTM models may be desirable because they can remember values over arbitrary time intervals. This, in turn, may allow for classifying events in input time series data for particular entities, as there may be lags of unknown duration between important events in the time series. However, other types of ML models are consistent with the disclosure, as are algorithmic models.
[0059] The present invention may be a system, a method, and/or a computer program product at any possible technical detail level of integration. The computer program product may include a computer readable storage medium (or media) having computer readable program instructions thereon for causing a processor to carry out aspects of the present invention.
[0060] The computer readable storage medium can be a tangible device that can retain and store instructions for use by an instruction execution device. The computer readable storage medium may be, for example, but is not limited to, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. A non-exhaustive list of more specific examples of the computer readable storage medium includes the following: a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD- ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanically encoded device such as punch-cards or raised structures in a groove having instructions recorded thereon, and any suitable combination of the foregoing. A computer readable storage medium, as used herein, is not to be construed as being transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide or other transmission media (e.g., light pulses passing through a fiber-optic cable), or electrical signals transmitted through a wire.
[0061] Computer readable program instructions described herein can be downloaded to respective computing/processing devices from a computer readable storage medium or to an external computer or external storage device via a network, for example, the Internet, a local area network, a wide area network and/or a wireless network. The network may comprise copper transmission cables, optical transmission fibers, wireless transmission, routers, firewalls, switches, gateway computers and/or edge servers. A network adapter card or network interface in each computing/processing device receives computer readable program instructions from the network and forwards the computer readable program instructions for storage in a computer readable storage medium within the respective computing/processing device.
[0062] Computer readable program instructions for carrying out operations of the present invention may be assembler instructions, instruction-set-architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, state-setting data, configuration data for integrated circuitry, or either source code or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, C++, or the like, and procedural programming languages, such as the "C" programming language or similar programming languages. The computer readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet Service Provider). In some embodiments, electronic circuitry including, for example, programmable logic circuitry, field-programmable gate arrays (FPGA), or programmable logic arrays (PLA) may execute the computer readable program instructions by utilizing state information of the computer readable program instructions to personalize the electronic circuitry, in order to perform aspects of the present invention.
[0063] Aspects of the present invention are described herein with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and/or block diagrams, and combinations of blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer readable program instructions.
[0064] These computer readable program instructions may be provided to a processor of a computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks. These computer readable program instructions may also be stored in a computer readable storage medium that can direct a computer, a programmable data processing apparatus, and/or other devices to function in a particular manner, such that the computer readable storage medium having instructions stored therein comprises an article of manufacture including instructions which implement aspects of the function/act specified in the flowchart and/or block diagram block or blocks.
[0065] The computer readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other device to produce a computer implemented process, such that the instructions which execute on the computer, other programmable apparatus, or other device implement the functions/acts specified in the flowchart and/or block diagram block or blocks.
[0066] The flowchart and block diagrams in the Figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the blocks may occur out of the order noted in the Figures. For example, two blocks shown in succession may, in fact, be accomplished as one step, executed concurrently, substantially concurrently, in a partially or wholly temporally overlapping manner, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems that perform the specified functions or acts or carry out combinations of special purpose hardware and computer instructions.
[0067] Embodiments of the present invention may also be delivered as part of a service engagement with a client corporation, nonprofit organization, government entity, internal organizational structure, or the like. Aspects of these embodiments may include configuring a computer system to perform, and deploying software and web services that implement, some or all of the methods described herein. Aspects of these embodiments may also include analyzing the client's operations, creating recommendations responsive to the analysis, generating software to implement portions of the recommendations, integrating the software into existing processes and infrastructure, metering use of the systems, allocating expenses to users of the systems, and billing for the use of the systems. This service engagement may be directed at providing both the cloud services and the cloud controller services, may be limited to only providing cloud controller services, or some combination thereof. Accordingly, these embodiments may further comprise receiving billing information from other entities and associating that billing information with end-users of the cloud.
[0068] The descriptions of the various embodiments of the present disclosure have been presented for purposes of illustration, but are not intended to be exhaustive or limited to the embodiments disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The terminology used herein was chosen to explain the principles of the embodiments, the practical application or technical improvement over technologies found in the marketplace, or to enable others of ordinary skill in the art to understand the embodiments disclosed herein.
[0069] Therefore, it is desired that the embodiments described herein be considered in all respects as illustrative, not restrictive, and that reference be made to the appended claims for determining the scope of the invention.
Furthermore, as an ordered combination, these elements amount to generic computer components receiving or transmitting data over a network, performing repetitive calculations, electronic record keeping, and storing and retrieving information in memory, which, as held by the courts, are well-understood, routine, and conventional. See MPEP 2106.05(d).
Moreover, the remaining elements of dependent claims do not transform the recited abstract idea into a patent eligible invention because these remaining elements merely recite further abstract limitations that provide nothing more than simply a narrowing of the abstract idea recited in the independent claims.
Looking at these limitations as an ordered combination adds nothing additional that is sufficient to amount to significantly more than the recited abstract idea because they simply provide instructions to use a generic arrangement of generic computer components to “apply” the recited abstract idea, perform insignificant extra-solution activity, and generally link the abstract idea to a technical environment. Thus, the elements of the claims, considered both individually and as an ordered combination, are not sufficient to ensure that the claim as a whole amounts to significantly more than the abstract idea itself. Since there are no limitations in these claims that transform the exception into a patent eligible application such that these claims amount to significantly more than the exception itself, claims 21-40 are rejected under 35 U.S.C. 101 as being directed to non-statutory subject matter.
Claim Rejections – 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries set forth in Graham v. John Deere Co., 383 U.S. 1, 148 USPQ 459 (1966), that are applied for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
Determining the scope and contents of the prior art.
Ascertaining the differences between the prior art and the claims at issue.
Resolving the level of ordinary skill in the pertinent art.
Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 21-40 is/are rejected under 35 U.S.C. 103 as being unpatentable by JP Patent Publication to US20200372349A1 to Ospina et al., (hereinafter referred to as “Ospina”) in view of US Patent Publication to US20200359550A1 to Tran et al., (hereinafter referred to as “Tran”)
As per Claim 21, Ospina teaches: (Currently Amended) A computer-based method for automatically identifying significant events in an environment, the method comprising: ([0069])
during a training phase: ([0146])
training a neural network to automatically identify significant events in the environment, wherein the neural network comprises a plurality of artificial cells interconnected via a plurality of gates, and wherein each gate encodes a strength of a relationship in the connection between an output of one artificial cell and an input of another artificial cell, the training comprising: (in at least [0098]-[0107][0074] in machine learning, a neural network needs to be trained and validated on labeled data, and the amount of training data required depends on the complexity of the problem as well as the complexity of the neural network. CLIMATEAI system 210 exploits the forecasting potential of simulated global climate data such as 202, 204, 206, and 208 in the training process.)
receiving training data that comprises labeled weather event data for a plurality of locations over a plurality of time periods; (in at least [0100] NN 500 may be setup to take, as input, a 24-month time series of monthly surface temperatures on a global 192×96 map or grid, and to forecast the Nino-3.4 index at a specified target lead-time such as 6 months into the future. FIG. 5B shows a graphical representation 550 of an exemplary data input, such as input climate data image 510, into NN 500, according to some embodiments of the present invention. In FIG. 5B, land surface and sea surface temperature anomalies on Apr. 1, 1980 are plotted on a two-dimensional grid 550 on the Earth's surface. This two-dimensional grid 550 in FIG. 5B is of size 26×14, much smaller than 192×96, for illustrative purpose only. [0107] Training data 625 is a documented dataset containing multiple instances of system inputs (e.g., input climate variables) and correct outcomes (e.g., forecasting results of output climate variables). It trains the ML model to optimize the performance for a specific target task, such as forecasting a specific target output climate variable at a specific target lead-time. In diagram 600, training data 625 may also include subsets for validating and testing the ML model. For an NN-based ML model, the quality of the output may depend on (a) NN architecture design and hyperparameter configurations, (b) NN coefficient or parameter optimization, and (c) quality of the training data set. These components may be refined and optimized using various methods. For example, training data 625 may be expanded via a climate data augmentation process. [0111] six climate simulation datasets 711, 712, 713, 714, 715, and 716 are first analyzed based on forecast targets 718. For example, simulation datasets CNRM-CM5, MPI-ESM-LR, GISS-E2-H, NorESM1-M, HadGEM2-ES, and GFDL-ESM2G may be used. For simplicity, it is assumed that the six datasets are from six different GCMs here. It would thus be understood that a “GCM” here also refers to the corresponding simulation dataset directly. In some embodiments, some of the datasets 711 to 716 may be from the same GCM, but were generated under different model parameters and/or initial conditions. In addition, only six datasets are drawn in FIG. 7 for illustrative purposes only. In another exemplary embodiment, forty input GCMs or GCM datasets may be used as input to the process, with ten validated via step 720, and six identified as skillful via step 740.)
generating, by the neural network, a probability factor representing a chance that a weather event in the training data will be significant for at least one location amongst the plurality of locations; and (in at least [0106] The training process begins at step 610 with data acquisition, retrieval, assimilation, or generation. At step 620, acquired data are pre-processed, or prepared. At step 630, the ML model is trained using training data 625. At step 640, the ML model is evaluated, validated, and tested, and further refinements to the ML model are fed back into step 630 for additional training. Once its performance is acceptable, at step 650, optimal model parameters are selected, for deployment at step 660. New data 655 may be used by the deployed model to make predictions. [0113] A step 720, a validation measure is computed for each GCM or GCM dataset, for at least one climate variable using observational historical climate data 716. For example, a GCM may be validated around one or more signal statistics of the at least one climate variable, where the validation measure measures the closeness between GCM simulation data statistics and that of observational historical data. In some embodiments, the at least one climate variable may be the target output climate variable as specified by forecasting targets 718 for a specific forecasting application. In some embodiments, the at least one climate variable may be one or more climate variables having direct functional dependencies with the target output climate variable. In yet some embodiments, the at least one climate variable may be an input climate variable to the forecasting system, one or more climate variables having direct functional dependencies with the input climate variable, or any other set of climate variables having some measurable significance on the target forecasting application. In one instance, the at least one climate variable is sea surface temperature for an El Nino region, and the statistics of interest may include bias, variance, correlation, autocorrelation over time, frequency of peaks, and the like. Sea surface temperature may be chosen here based on the target output variable being the Nino-3.4 index. [0114] GCM simulation datasets for which signal statistics match, or is close to the same statistics of the observational historical data may be considered a properly modeled, or validated dataset. In some embodiments, a computed GCM data statistic may be referred to as the validation measure, while a corresponding computed observational historical data statistic may be referred to as the validation threshold. In some embodiments, the validation measure may be a measure of closeness as computed through Mahalanobis distance, Euclidean distance, or any other appropriate distance measure between the GCM simulation data and the observational historical data, between a GCM simulation data statistic and the corresponding observational historical data statistic, or between several GCM simulation data statistics and the corresponding observational historical data statistics. A validated dataset may have a validation measure below a static or dynamic validation threshold. For example, GCMs may be ranked based on their validation measures, and the validation threshold may be chosen so a specific number of GCMs are considered validated. In some embodiments that uses the Mahalanobis distance, climate variables of interest may be transformed into uncorrelated variables first with their variances scaled to 1, prior to calculating the distance measures. In this illustrative example, four GCM datasets 711, 712, 713, and 716 have acceptable validation measures and make up a validated GCM subset 730. [0145] land-sea masks may be used for occluding selected portions of the climate data image to focus on land only or on sea only, for example when the land is a poor predictor for the desired climate forecast. In an illustrative example, 3-10 augmented climate data images may be generated by occluding random-sized boxes having 10% to 30% of the total area, where data within the random-sized boxes are set to 0. Such an image occlusion process may be run probabilistically, as configured through a hyperparameter, set to a value within a given range, such as between 0.1 and 0.9. The probability of re-running the occlusion process on the same image may also be configured to a value within a given range, such as between 0.1 and 0.3.)
comparing the generated probability to the labeled weather event data of the received training data; and (in at least [0065] the CLIMATEAI climate forecasting system employs a deep learning network that is capable of extracting spatial-temporal features as well as functional dependencies and correlations among different GCM simulation datasets to predict future climate conditions. Typically in supervised learning, a predictor model such as a neural network is first trained using a first set of labeled training data to determine an optimal set of internal parameters. The capability of the predictor model is then validated on a second validation dataset and tuned accordingly. A third test dataset is then used to evaluate the predictive or forecast skill of the model. [0067] The first is a process to validate the forecasting potential of individual GCMs developed by different agencies under different assumptions, and to generate a multi-model data ensemble by selecting and combining multiple validated and skillful GCMs or GCM datasets. Such selection and combination of GCM datasets are performed in successive stages, possibly with multiple iterations or passes, to minimize computation overheads without significantly compromising accuracy of the end result. [0114] GCM simulation datasets for which signal statistics match, or is close to the same statistics of the observational historical data may be considered a properly modeled, or validated dataset. In some embodiments, a computed GCM data statistic may be referred to as the validation measure, while a corresponding computed observational historical data statistic may be referred to as the validation threshold. In some embodiments, the validation measure may be a measure of closeness as computed through Mahalanobis distance, Euclidean distance, or any other appropriate distance measure between the GCM simulation data and the observational historical data, between a GCM simulation data statistic and the corresponding observational historical data statistic, or between several GCM simulation data statistics and the corresponding observational historical data statistics. A validated dataset may have a validation measure below a static or dynamic validation threshold. For example, GCMs may be ranked based on their validation measures, and the validation threshold may be chosen so a specific number of GCMs are considered validated. In some embodiments that uses the Mahalanobis distance, climate variables of interest may be transformed into uncorrelated variables first with their variances scaled to 1, prior to calculating the distance measures. In this illustrative example, four GCM datasets 711, 712, 713, and 716 have acceptable validation measures and make up a validated GCM subset 730. [0115] At step 740, validated GCMs or GCM simulation datasets are further evaluated for their ability to forecast the target output climate variable, or for their ability to forecast some climate variables highly correlated with the target output climate variable. In some embodiments, a forecast skill score is computed for each validated GCM based on a forecast function, where the forecast function may be a data predictor function, the NN-based climate forecasting model which will be trained using data ensemble 790, or a model-analog. [0135] Statistical augmentation broadly refers to generating synthetic but feasible states of climate variables, such as temperature and precipitation, that have the same statistical distributions as real observational climate data or GCM simulation data. One statistical augmentation method is the McKinnon Data Augmentation (MDA) technique, which leverages three climate indices: El Nino-Southern Oscillation (ENSO), Atlantic Multidecadal Oscillation (AMO), and Pacific Decadal Oscillation (PDO). These indices measure the state of the warm and cold temperature cycles in the Equatorial Pacific Ocean, the Atlantic Ocean, and the Northern Pacific Ocean, and affect climate variables such as temperature and precipitation throughout the world. MDA may also extract a global forcing signal (e.g., global warming), and an internal variability signal (e.g., weather). MDA first uses linear regression to learn the influences of ENSO, AMO, and PDO on a climate variable at each location, then represents the value of the climate variable as the sum of six components: the mean value of the variable, the influences of ENSO, the influence of AMO, the influence of PDO, the influence of climate change, and internal climate variability. MDA further uses surrogate time series generation techniques to generate alternate ENSO, AMO and PDO time series with means, variances, and autocorrelations similar to the original time series, and bootstraps to generate alternate estimates of internal climate variability. MDA then substitutes in new values of the influences of ENSO, AMO, and PDO, and the effect of internal climate variability into its representation of the climate system as the sum of different components. This process is repeated for each location on earth multiple times to generate alternate viable climate states, which may in turn be used for GCM data augmentation.)
in response to the comparison, updating the encoded strength of at least some of the plurality of gates of the neural network; (in at least [0099] FIG. 5A is an exemplary artificial neural network design for climate forecasting, according to some embodiments of the present invention. This exemplary neural network (NN) 500 is for illustration only and does not limit the scope of the invention to the particular NN architecture and particular forecasting application shown. NNs can be viewed as parallel, densely interconnected computational models that adaptively learn through automatic adjustment of system parameters based on training data. Input information are modified based on system parameters when traversing through layers of interconnected neurons or nodes, to activate or trigger particular outputs. The design of an NN refers to the configuration of its architecture and topology, or the specific arrangements of layers and nodes in the network. In some embodiments, the design of the NN may also comprise determination or configuration techniques for pre-training the NN and/or for initialization of hyperparameters and model coefficients. [0104] FIG. 5C is an exemplary Long Short-Term Memory (LSTM) cell 580 for use in RNN 530 in FIG. 5A. LSTMs are a special type of RNN capable of learning long-term dependencies in sequence prediction problems. The long-term memory refers to learned weights, and the short-term memory refers to gated cell states. [0116] The data predictor function may predict the target output variable such as the Nino-3.4 index directly from a GCM dataset, or may predict another closely related climate variable, such as the Nino-3 index and/or the Nino-4 index. This data predictor function may be viewed as a coarse filter to determine whether a given GCM dataset has any forecast power towards the target output climate variable, before more complex filters or metrics are applied. [0117] Evaluating forecasting skills of a GCM using the NN-based climate forecasting model directly may be viewed as a recursive or iterative approach, where each GCM is individually assessed preliminarily, before being augmented further into an ensemble, or before multiple GCMs are combined into an ensemble. The resulting data ensemble may be further assessed for its overall forecasting skills. Here the NN-based climate forecasting model may have been pre-trained, for example on limited amount of reanalysis data, or pre-existing image recognition databases. [0119] Forecast skills may be measured by a mean square error (MSE), a correlation between the forecast and the actual values of the target climate variable, or other appropriate error or distance metrics. Such an MSE or correlation value computed for the forecast function discussed above may be viewed as a forecast skill score, and used for selecting a validated and skillful subset of GCMs. Depending on the definition of the forecast skill score, a best-scored GCM may be one with a high forecast skill score, or one with a low forecast skill score. In this illustrative example, three GCM datasets 711, 712, and 716 are chosen as best-scored datasets and make up a validated and skillful GCM subset 750. Again, forecast skill scores may be compared to a threshold; GCM datasets may also be ranked based on their forecast skill scores and a desired number of GCMs may be selected. [0131] spatial re-gridding may be performed iteratively, where the forecast skill of a spatially re-gridded multi-model GCM simulation data ensemble is evaluated, and the common spatial scale modified, updated, or entirely regenerated based on the forecast skill. In some embodiments, such iterative spatial re-gridding may occur subsequent to step 740 or 760 in FIG. 7. [0161] Table 2 shows rankings of different GCM datasets and different ensembles of GCM datasets in predicting AIR and SKT, as measured by Root Mean Square Error (RMSE) when compared to ground truth Nino-3.4 surface temperature targets calculated from the ECMWF ERAS dataset, which is a gridded dataset of reanalyzed historical observations. In Table 2, “ii+tr(i)” refers to a data ensemble generated by concatenating models ii and i, and “i+tr(ii)” refers to a concatenation of the two models in a reverse order. A concatenation ii+tr(i) of models ii and i is the process of feeding model ii into the NN for training first, followed by model i; a concatenation i+tr(ii) is the process of feeding model i into the NN for training first, followed by model ii. The notation “tr( )” is used to represent a transfer learning process where different models i and ii are both used to train the same NN for a climate forecasting application. In addition, “[ii+tr(i)]×100” refers to repeating “ii+tr(i)” one hundred times as a data augmentation measure. Using such an augmented dataset is equivalent to running or training the forecasting model one hundred times on the same ii+tr(i) data concatenation or ensemble, but each time with different hyperparameters and weight initializations as derived from the previous run. The forecasting model output at the end of the 100 respective runs may also be viewed collectively and probabilistically as a conventional ensemble forecasting distribution.)
during a runtime phase: ([0146][0151])
receiving data from a plurality of …, wherein each of the … is configured to generate weather event data about a respective one or more locations, the weather event data comprising environmental measurements for inferring significance of weather events at the one or more locations; (in at least [0061] a GCM is initialized with observed or estimated atmosphere, ocean, land, and sea ice states, and run in forward time into the future. Because of the chaotic nature of the climate system, forecast results can be very sensitive to even small perturbations to the initial conditions or model parameters of the system. Any changes in perturbations or external forcings to the system, for example in the form of solar irradiance, or human contributed carbon and aerosol emissions, would require a GCM-based forecast to be run again, and any additional lead-time for the forecast requires at least polynomial increase in the total number of computations needed, while also increasing the amount of forecast uncertainty. Moreover, differences in GCM model design often lead to very different forecasting skills, with some models performing better than others in some specific climate forecasting applications. Models can also perform better at some specific time of the year than at other times of the year. Ensemble modeling such as seasonal predictions through the North American Multi-Model Ensemble (NMME) has been studied to reduce forecast uncertainty by post-processing, ranking, weighting, and averaging climate projections from different GCMs, yet such approaches may require even higher computational power [0063] machine-learning based climate forecast systems have utilized long short-term memory neural networks or a combination of autoregressive integrated moving average models and artificial neural networks. Such forecasts are trained and validated exclusively on observational historical data, and are thus significantly constrained by the short observational record of climate data, which has only been measured in situ or via satellites on the global scale for the past hundred years or so. [0113] A step 720, a validation measure is computed for each GCM or GCM dataset, for at least one climate variable using observational historical climate data 716. For example, a GCM may be validated around one or more signal statistics of the at least one climate variable, where the validation measure measures the closeness between GCM simulation data statistics and that of observational historical data. In some embodiments, the at least one climate variable may be the target output climate variable as specified by forecasting targets 718 for a specific forecasting application. In some embodiments, the at least one climate variable may be one or more climate variables having direct functional dependencies with the target output climate variable. In yet some embodiments, the at least one climate variable may be an input climate variable to the forecasting system, one or more climate variables having direct functional dependencies with the input climate variable, or any other set of climate variables having some measurable significance on the target forecasting application. In one instance, the at least one climate variable is sea surface temperature for an El Nino region, and the statistics of interest may include bias, variance, correlation, autocorrelation over time, frequency of peaks, and the like. Sea surface temperature may be chosen here based on the target output variable being the Nino-3.4 index. [0133] each of spatial re-gridding and temporal homogenization may be performed on a subset of the GCM datasets, or be performed on different subsets at different stages, before or after the generation of the multi-model ensemble. For example, optional steps 765 and 775 in FIG. 7 illustrate two different locations in the GCM model selection and data ensemble generation process, where data homogenization may be carried out. [0168] The GUI may illustrate historical wind speed analysis along with historical AI-based reconstruction that allows review of 100 years of data and calculating a true mean. Based on the analysis, a user may then determine that there is a large difference between reconstructed mean and historical mean (most of the time in some locations). The detailed analysis allows making extrapolations that reduce the margin of uncertainty. Knowing the correct projection can help in wind resource planning and correctly incorporating climate risks into business decisions.)
providing the weather event data as inputs to the trained neural network; (in at least [0148] based on the specific forecasting application as specified by a desired output climate variable, an input variable to the NN may be customized. For example, to predict surface temperature, 2 m air temperature or air temperature at 2 meters above the surface may be used as input; to predict sea surface temperate, air temperature may be used as input. Moreover, depending on the target lead-time, and/or a specific time horizon which is a fixed point of time in the future when the forecast occurs, the structure or architecture of the NN may be customized. For example, to forecast January sea surface temperature at 9-months lead-time, the NN may be customized to forecast month-by-month to predict Januaries 9-months ahead. [0151] Once trained, the climate forecasting NN may be validated and tested at step 950, using validation and test sets comprising observational historical data. In some embodiments, all available observational historical data including reanalysis data may be divided by year into a tuning set, a validation set, and/or a testing set, for use in steps 940 and 950. Based on validation results that indicate forecast uncertainty or confidence, any of the steps 910, 920, 930 and 940 may be repeated to update the NN climate forecasting model and improve performance of the forecasting system. The fully trained deployed in step 950 for actual climate forecasting.)
receiving, as output from the trained neural network, weather event data having a respective significance level that is greater than a respective predetermined threshold value; and (in at least [0113] the at least one climate variable may be an input climate variable to the forecasting system, one or more climate variables having direct functional dependencies with the input climate variable, or any other set of climate variables having some measurable significance on the target forecasting application. In one instance, the at least one climate variable is sea surface temperature for an El Nino region, and the statistics of interest may include bias, variance, correlation, autocorrelation over time, frequency of peaks, and the like. Sea surface temperature may be chosen here based on the target output variable being the Nino-3.4 index. [0150] Further fine-tuning of the NN may occur at step 940, based on observational historical data 935 including reanalysis data 938. Recall that climate reanalysis combines and assimilates observational historical data with physical dynamical models to “fill in gaps” and provide a physically coherent and consistent, synthesized estimate of the climate in the past, while keeping the historical record uninfluenced by artificial factors. The availability of reanalysis data is limited by that of the observational historical data, and typical reanalysis datasets span over 40 to 80 years. To maximize the effectiveness of reanalysis data 938 in step 940, in some embodiments, some NN layers may be frozen during tuning, with only a selected subset of NN layers further updated. For example, in CRNN 500 shown in FIG. 5A, the first five convolutional layers and the RNN may be frozen, while the last convolutional layer is fine-tuned on 80 years of reanalysis data. [0151] Once trained, the climate forecasting NN may be validated and tested at step 950, using validation and test sets comprising observational historical data. In some embodiments, all available observational historical data including reanalysis data may be divided by year into a tuning set, a validation set, and/or a testing set, for use in steps 940 and 950. Based on validation results that indicate forecast uncertainty or confidence, any of the steps 910, 920, 930 and 940 may be repeated to update the NN climate forecasting model and improve performance of the forecasting system. The fully trained deployed in step 950 for actual climate forecasting. [0153] FIG. 11 is another exemplary flow diagram 1100 for a process to generate a multi-model ensemble of global climate simulation data, according to some embodiments of the present invention. The process flow shown in diagram 1100 is an exemplary implementation of step 1010 in FIG. 10. Upon initialization at step 1105, a plurality of Global Climate Models (GCMs) are first examined. A GCM validation measure is computed for each of the plurality of GCMs, based on at least one sample statistic for at least one climate variable of simulation data from the GCM. At step 1120, a validated subset of the plurality of GCMs is selected, by comparing each computed GCM validation measure to a validation threshold determined based on a set of observational historical climate data. At step 1130, a forecast skill score is computed for each validated GCM, based on a forecast function selected from the group consisting of a first data predictor function, the NN-based climate forecasting model, and a model analog of the NN-based climate forecasting model. Next at step 1140, a validated and skillful subset of GCMs is selected by choosing at least two best-scored GCMs. At step 1150, one or more candidate ensembles of global climate simulation data are generated by combining simulation data from at least two validated and skillful GCMs from the validated and skillful subset of GCMs. At step 1160, an ensemble forecast skill score is computed for each candidate ensemble, based on a forecast function selected from the group consisting of a second ensemble-based data predictor function, the NN-based climate forecasting model, and a model analog of the NN-based climate forecasting model. At step 1170, a multi-model ensemble of global climate simulation data is generated by selecting a best-scored candidate ensemble of global climate simulation data. The overall process terminates at step 1180.)
correlating weather events with the outputted weather event data having the respective significant level that is greater than the respective predetermined threshold value. (in at least [0065] the CLIMATEAI climate forecasting system employs a deep learning network that is capable of extracting spatial-temporal features as well as functional dependencies and correlations among different GCM simulation datasets to predict future climate conditions. Typically in supervised learning, a predictor model such as a neural network is first trained using a first set of labeled training data to determine an optimal set of internal parameters. The capability of the predictor model is then validated on a second validation dataset and tuned accordingly. A third test dataset is then used to evaluate the predictive or forecast skill of the model. [0115] At step 740, validated GCMs or GCM simulation datasets are further evaluated for their ability to forecast the target output climate variable, or for their ability to forecast some climate variables highly correlated with the target output climate variable. In some embodiments, a forecast skill score is computed for each validated GCM based on a forecast function, where the forecast function may be a data predictor function, the NN-based climate forecasting model which will be trained using data ensemble 790, or a model-analog. [0164] FIG. 13A is a graph 1300 comparing correlation measures of different forecasting methods in predicting the Nino 3.4 Index at different lead times, according to some embodiments of the present invention. Meanwhile, FIG. 13B is a graph 1350 comparing the time series of forecast results by different forecasting methods in predicting the Nino 3.4 Index, according to some embodiments of the present invention. The CLIMATEAI forecast result shown in FIGS. 13A and 13B are based on an NN trained on a data ensemble that interleaves simulation data from multiple GCMs. It can be seen that neural networks trained on AOGCMs according to embodiments of the present invention offer comparable forecasting performance to SEAS5 while consuming only a miniscule fraction of the computation power. The CLIMATEAI system that learns from abundant AOGCM simulations also outperforms NNs trained purely on limited amount of historical observations. [0165] the CLIMATEAI system may be trained according to embodiments of the present invention to forecast some primary climate variable such as surface temperature, and an additional shallow neural network may be connected to the NN output to further predict some secondary climate variables such as wind speed, streamflow, growing degree days (GDD), daily temperature [0169] FIG. 16 is an illustrative diagram 1600 showing an analysis of power generation by a hydroelectric plant including seasonal variations in power production, according to some embodiments of the present invention. Total generations and river stream flows are compared in the bottom panel, indicating a high correlation with seasonality in power generation at this site. That is, FIG. 15 illustrates yet another GUI that can be accessed by a user to access or operate a system implementing some embodiments of the present invention. [0172] the user selection includes an industry and one or more parameters associated with the industry. In some embodiments, the industry is an energy industry, a real estate industry, an agricultural industry, a financial industry, and/or an insurance industry. In some embodiments, the energy industry includes a wind power industry, and the one or more parameters include one or more of a location, a wind power farm, and a forecast horizon. In some embodiments, the energy industry includes a hydroelectric industry, and the one or more parameters include one or more of a location, a hydroelectric dam, and a forecast horizon.)
Although implied, Ospina does not expressly disclose the following limitations, which however, are taught by Tran,
receiving data from a plurality of sensors, wherein each of the sensors is configured to generate… (in at least [0014] a motorized frame to move in a field, cameras, temperature sensors, and other sensors to capture field data, and machine learning to monitor and manage problems in field. One implementation can collect weather data like hail, wind, rainfall, and temperature, which can be placed in fields to assist in data-driven field decisions. One embodiment supplements local weather data with external sources like satellites and radar. This embodiment can correct for what is being provided in the public realm through satellites and radar—making it more precise and accurate—by fusing it with the insight gleaned from irrigation ground-based stations. [0065] to monitor weather, another system includes a motorized frame to move in a field, cameras, temperature sensors, and other sensors to capture field data, and machine learning to monitor and manage problems in field. One implementation can collect weather data like hail, wind, rainfall, and temperature, which can be placed in fields to assist in data-driven field decisions. One embodiment supplements local weather data with external sources like satellites and radar. This embodiment can correct for what is being provided in the public realm through satellites and radar—making it more precise and accurate—by fusing it with the insight gleaned from irrigation ground-based stations. [0189] Other in- or on-ground sensors can be deployed to detect crop conditions, weather data, and many other details, which can then be transmitted to decision analytics platforms via the Internet of Things (where computing devices embedded in everyday objects are connected to the Internet to enable analytics). A solar-powered in-ground sensor can gather data on crop stress, air pressure, humidity, temperature, chlorophyll, canopy biomass, rainfall, and other information, which can then be analyzed on its platform to improve precision farming.)
At the time the invention was filed, it would have been obvious for one of ordinary skill in the art to have modified the teachings of Ospina, as taught by Tran above, with a reasonable expectation of success if arriving at the claimed invention. One of ordinary skill in the art would have been motivated to make this modification to the teachings of Ospina with the motivation of, …improved irrigation scheduling and efficiency. While soil moisture data (from either sensors or models) have long been used as a scheduling aid, AI provides machine learning of how soil moisture responds to irrigation events in scenarios with different crops, soils, environmental conditions, etc. Tied to an irrigation control system, the AI machine can automatically implement control strategies that help minimize water usage, manage nutrient losses, or achieve more desirable or uniform soil moisture throughout the field....the data collected during the planting and growing seasons provide for visibility into the ag supply chain and continued improvement of the core microbiome solutions…A solar-powered in-ground sensor can gather data on crop stress, air pressure, humidity, temperature, chlorophyll, canopy biomass, rainfall, and other information, which can then be analyzed on its platform to improve precision farming...These reports may include recommendations for changing the inputs to the farm 12 to improve or alter the quality of the outputs according to particular goals of the farm 12 or the consumers 16...precision monitoring technology, historic weather data, and other sources are learned by artificial intelligence/learning machines to conduct detailed analysis, ranging from the descriptive to the prescriptive. By improving the volume, quality, flow, and frequency of information used in farming and other value chain stages, food production can become more efficient, productive, and sustainable. The volume of data for agribusiness is steadily expanding due to sensors, satellite monitoring, and other information gathering technologies. The quality of information is improving with more sophisticated data-gathering instruments, as well as crowd-sourced farmer data whose privacy is secured through the blockchain. The flow of information is enhanced via platforms that connect various stakeholders across the value chain. Finally, the frequency of information is increasing with improved Internet connectivity, device-enabled products sending data to cloud analytics platforms (i.e., Internet of Things), and many other advancements. As the quantity, quality, speed, and flow of data improves, data analytics platforms and machine learning applications can enable better practices in farming, processing, and manufacturing…uses AI for improved irrigation scheduling and efficiency. While soil moisture data (from either sensors or models) have long been used as a scheduling aid, A provides machine learning of how soil moisture responds to irrigation events in scenarios with different crops, soils, environmental conditions, etc. Tied to an irrigation control system, the A machine can automatically implement control strategies that help minimize water usage, manage nutrient losses, or achieve more desirable or uniform soil moisture throughout the field...to provide a user of the plant counting mobile irrigation system a way to evaluate the resulting data and establish a higher level of confidence in the accuracy of such data…AI learns the associations between available weather, crop and soil condition data, and the corresponding irrigation recommendations of a trained agronomist, thereby automating the repetitive aspects of the scheduling process…The irrigation system may also adapt the driving to the observed environmental changes such as weather or farm conditions…, as recited in Tran.
As per Claim 22, Ospina teaches: (Currently Amended) The method of claim 21, wherein based on receiving, as output from the trained neural network, weather events having respective significance levels that are greater than predetermined threshold values value for the weather events, the method further comprises:
calculating a probability, using the trained neural network, that a correlated weather event will be significant for the respective one or more locations based on the respective significance levels; and (in at least [0113] A step 720, a validation measure is computed for each GCM or GCM dataset, for at least one climate variable using observational historical climate data 716. For example, a GCM may be validated around one or more signal statistics of the at least one climate variable, where the validation measure measures the closeness between GCM simulation data statistics and that of observational historical data. In some embodiments, the at least one climate variable may be the target output climate variable as specified by forecasting targets 718 for a specific forecasting application. In some embodiments, the at least one climate variable may be one or more climate variables having direct functional dependencies with the target output climate variable. In yet some embodiments, the at least one climate variable may be an input climate variable to the forecasting system, one or more climate variables having direct functional dependencies with the input climate variable, or any other set of climate variables having some measurable significance on the target forecasting application. In one instance, the at least one climate variable is sea surface temperature for an El Nino region, and the statistics of interest may include bias, variance, correlation, autocorrelation over time, frequency of peaks, and the like. Sea surface temperature may be chosen here based on the target output variable being the Nino-3.4 index. [0114] GCM simulation datasets for which signal statistics match, or is close to the same statistics of the observational historical data may be considered a properly modeled, or validated dataset. In some embodiments, a computed GCM data statistic may be referred to as the validation measure, while a corresponding computed observational historical data statistic may be referred to as the validation threshold. In some embodiments, the validation measure may be a measure of closeness as computed through Mahalanobis distance, Euclidean distance, or any other appropriate distance measure between the GCM simulation data and the observational historical data, between a GCM simulation data statistic and the corresponding observational historical data statistic, or between several GCM simulation data statistics and the corresponding observational historical data statistics. A validated dataset may have a validation measure below a static or dynamic validation threshold. For example, GCMs may be ranked based on their validation measures, and the validation threshold may be chosen so a specific number of GCMs are considered validated. In some embodiments that uses the Mahalanobis distance, climate variables of interest may be transformed into uncorrelated variables first with their variances scaled to 1, prior to calculating the distance measures. In this illustrative example, four GCM datasets 711, 712, 713, and 716 have acceptable validation measures and make up a validated GCM subset 730. [0115] At step 740, validated GCMs or GCM simulation datasets are further evaluated for their ability to forecast the target output climate variable, or for their ability to forecast some climate variables highly correlated with the target output climate variable. In some embodiments, a forecast skill score is computed for each validated GCM based on a forecast function, where the forecast function may be a data predictor function, the NN-based climate forecasting model which will be trained using data ensemble 790, or a model-analog. [0119] a forecasting model may be considered “skillful” if it can better predict a target output climate variable than a random guess, a historical average, or an average value computed from GCM data. Forecast skills may be measured by a mean square error (MSE), a correlation between the forecast and the actual values of the target climate variable, or other appropriate error or distance metrics. Such an MSE or correlation value computed for the forecast function discussed above may be viewed as a forecast skill score, and used for selecting a validated and skillful subset of GCMs. Depending on the definition of the forecast skill score, a best-scored GCM may be one with a high forecast skill score, or one with a low forecast skill score. In this illustrative example, three GCM datasets 711, 712, and 716 are chosen as best-scored datasets and make up a validated and skillful GCM subset 750. Again, forecast skill scores may be compared to a threshold; GCM datasets may also be ranked based on their forecast skill scores and a desired number of GCMs may be selected. [0135] MDA first uses linear regression to learn the influences of ENSO, AMO, and PDO on a climate variable at each location, then represents the value of the climate variable as the sum of six components: the mean value of the variable, the influences of ENSO, the influence of AMO, the influence of PDO, the influence of climate change, and internal climate variability. MDA further uses surrogate time series generation techniques to generate alternate ENSO, AMO and PDO time series with means, variances, and autocorrelations similar to the original time series, and bootstraps to generate alternate estimates of internal climate variability. MDA then substitutes in new values of the influences of ENSO, AMO, and PDO, and the effect of internal climate variability into its representation of the climate system as the sum of different components. This process is repeated for each location on earth multiple times to generate alternate viable climate states, which may in turn be used for GCM data augmentation. [0145] land-sea masks may be used for occluding selected portions of the climate data image to focus on land only or on sea only, for example when the land is a poor predictor for the desired climate forecast. In an illustrative example, 3-10 augmented climate data images may be generated by occluding random-sized boxes having 10% to 30% of the total area, where data within the random-sized boxes are set to 0. Such an image occlusion process may be run probabilistically, as configured through a hyperparameter, set to a value within a given range, such as between 0.1 and 0.9. The probability of re-running the occlusion process on the same image may also be configured to a value within a given range, such as between 0.1 and 0.3. [0148] a neural network such as CRNN 500 in FIG. 5 may be designed, generated, and/or updated, based on forecasting targets 905. That is, hyperparameters of CRNN 500 may be configured in step 925 for different forecasting applications. For example, a loss function for the NN may be chosen. For threshold applications, such as when an El Nino event is to be predicted, cross-entropy loss may be used; for numerical applications, such as when an average temperature is to be predicted, the sigmoid function may be used instead. Furthermore, based on the specific forecasting application as specified by a desired output climate variable, an input variable to the NN may be customized. For example, to predict surface temperature, 2 m air temperature or air temperature at 2 meters above the surface may be used as input; to predict sea surface temperate, air temperature may be used as input. Moreover, depending on the target lead-time, and/or a specific time horizon which is a fixed point of time in the future when the forecast occurs, the structure or architecture of the NN may be customized. For example, to forecast January sea surface temperature at 9-months lead-time, the NN may be customized to forecast month-by-month to predict Januaries 9-months ahead. [0150] Further fine-tuning of the NN may occur at step 940, based on observational historical data 935 including reanalysis data 938. Recall that climate reanalysis combines and assimilates observational historical data with physical dynamical models to “fill in gaps” and provide a physically coherent and consistent, synthesized estimate of the climate in the past, while keeping the historical record uninfluenced by artificial factors. The availability of reanalysis data is limited by that of the observational historical data, and typical reanalysis datasets span over 40 to 80 years. To maximize the effectiveness of reanalysis data 938 in step 940, in some embodiments, some NN layers may be frozen during tuning, with only a selected subset of NN layers further updated. For example, in CRNN 500 shown in FIG. 5A, the first five convolutional layers and the RNN may be frozen, while the last convolutional layer is fine-tuned on 80 years of reanalysis data. [0151] Once trained, the climate forecasting NN may be validated and tested at step 950, using validation and test sets comprising observational historical data. In some embodiments, all available observational historical data including reanalysis data may be divided by year into a tuning set, a validation set, and/or a testing set, for use in steps 940 and 950. Based on validation results that indicate forecast uncertainty or confidence, any of the steps 910, 920, 930 and 940 may be repeated to update the NN climate forecasting model and improve performance of the forecasting system. The fully trained deployed in step 950 for actual climate forecasting.)
returning information identifying the particular weather event based on the calculated probability. (in at least [0164] FIG. 13A is a graph 1300 comparing correlation measures of different forecasting methods in predicting the Nino 3.4 Index at different lead times, according to some embodiments of the present invention. Meanwhile, FIG. 13B is a graph 1350 comparing the time series of forecast results by different forecasting methods in predicting the Nino 3.4 Index, according to some embodiments of the present invention. The CLIMATEAI forecast result shown in FIGS. 13A and 13B are based on an NN trained on a data ensemble that interleaves simulation data from multiple GCMs. It can be seen that neural networks trained on AOGCMs according to embodiments of the present invention offer comparable forecasting performance to SEAS5 while consuming only a miniscule fraction of the computation power. The CLIMATEAI system that learns from abundant AOGCM simulations also outperforms NNs trained purely on limited amount of historical observations. [0167] FIG. 14 shows a diagram 1400 of a seasonal average wind speed prediction for the ASSURA II wind farm in Brazil, according to some embodiments of the present invention. That is, FIG. 14 illustrates a graphical user interface (GUI) that can be accessed or operated by a user to access a system implementing some embodiments of the present invention.)
As per Claim 23, Ospina teaches: (Currently Amended) The method of claim 22,
wherein the particular weather event is a next weather event in a series of weather events. (in at least [0061] When applied to climate forecasting, a GCM is initialized with observed or estimated atmosphere, ocean, land, and sea ice states, and run in forward time into the future. Because of the chaotic nature of the climate system, forecast results can be very sensitive to even small perturbations to the initial conditions or model parameters of the system. Any changes in perturbations or external forcings to the system, for example in the form of solar irradiance, or human contributed carbon and aerosol emissions, would require a GCM-based forecast to be run again, and any additional lead-time for the forecast requires at least polynomial increase in the total number of computations needed, while also increasing the amount of forecast uncertainty. Moreover, differences in GCM model design often lead to very different forecasting skills, with some models performing better than others in some specific climate forecasting applications. Models can also perform better at some specific time of the year than at other times of the year. Ensemble modeling such as seasonal predictions through the North American Multi-Model Ensemble (NMME) has been studied to reduce forecast uncertainty by post-processing, ranking, weighting, and averaging climate projections from different GCMs, yet such approaches may require even higher computational power. [0150] fine-tuning of the NN may occur at step 940, based on observational historical data 935 including reanalysis data 938. Recall that climate reanalysis combines and assimilates observational historical data with physical dynamical models to “fill in gaps” and provide a physically coherent and consistent, synthesized estimate of the climate in the past, while keeping the historical record uninfluenced by artificial factors. The availability of reanalysis data is limited by that of the observational historical data, and typical reanalysis datasets span over 40 to 80 years. To maximize the effectiveness of reanalysis data 938 in step 940, in some embodiments, some NN layers may be frozen during tuning, with only a selected subset of NN layers further updated. For example, in CRNN 500 shown in FIG. 5A, the first five convolutional layers and the RNN may be frozen, while the last convolutional layer is fine-tuned on 80 years of reanalysis data. [0164] FIG. 13A is a graph 1300 comparing correlation measures of different forecasting methods in predicting the Nino 3.4 Index at different lead times, according to some embodiments of the present invention. Meanwhile, FIG. 13B is a graph 1350 comparing the time series of forecast results by different forecasting methods in predicting the Nino 3.4 Index, according to some embodiments of the present invention. The CLIMATEAI forecast result shown in FIGS. 13A and 13B are based on an NN trained on a data ensemble that interleaves simulation data from multiple GCMs. It can be seen that neural networks trained on AOGCMs according to embodiments of the present invention offer comparable forecasting performance to SEAS5 while consuming only a miniscule fraction of the computation power. The CLIMATEAI system that learns from abundant AOGCM simulations also outperforms NNs trained purely on limited amount of historical observations. [0167] FIG. 14 shows a diagram 1400 of a seasonal average wind speed prediction for the ASSURA II wind farm in Brazil, according to some embodiments of the present invention. That is, FIG. 14 illustrates a graphical user interface (GUI) that can be accessed or operated by a user to access a system implementing some embodiments of the present invention.)
As per Claim 24, Ospina teaches: (Previously Presented) The method of claim 21,
wherein the training data comprises labeled seasonal activities and conditions data for the plurality of locations. (in at least [0157] The El Nino-Southern Oscillation (ENSO) is a cycle of warm (El Nino) and cold (La Nina) temperatures in the equatorial Pacific Ocean that influences weather patterns around the world. It impacts North American temperature and precipitation, the Indian Monsoon, and hurricanes in the Atlantic. Thus, it has consequences for agricultural planning, commodity prices, insurance terms, and energy availability. [0159] the CLIMATEAI system trains a NN 500 as shown in FIG. 5A on simulations from AOGCMs and evaluates the NN on observational historical data, to predict surface temperatures. More specifically, the CLIMATEAI system trains NN 500 on 24-month time series of monthly, preindustrial (piControl) surface temperature data from the following AOGCMs, each named after the modeling centers that produced them: CNRM-CM5 (800 years), MPI-ESM-LR (1000 years), NorESM1-M (500 years), HadGEM2-ES (500 years), and GFDL-ESM2G (500 years). The specific version numbers of the simulation datasets used for training are shown in Table 1. In Table 1, “tos” refers to Sea Surface Temperature, and “tas” refers to Near-Surface Air Temperature.)
As per Claim 25, Ospina teaches: (Previously Presented) The method of claim 21,
wherein the weather event data includes a weather forecast. (in at least [0152] FIG. 10 is another exemplary flow diagram 1000 for a process to generate and train an NN-based climate forecasting model, according to some embodiments of the present invention. Upon initialization at step 1005, global climate simulation data 1006 from at least two global climate simulation models are combined at step 1010 into a multi-model ensemble. The multi-model ensemble is then pre-processed at step 1020, where the pre-processing comprises at least one pre-processing action selected form the group consisting of spatial re-gridding, temporal homogenization, and data augmentation. At step 1030, a neural network (NN)-based climate forecasting model is trained on the pre-processed multi-model global climate simulation data ensemble, and validated on a set of observational historical climate data at step 1040. The observational historical climate data may comprise reanalysis data, and the NN's hyperparameters may be fine-tuned during the validation process as well. At an optional step 1050, the NN-based climate forecasting model may be deployed, and the process ends at step 1060.)
As per Claim 26, Ospina teaches: (Previously Presented) The method of claim 21,
wherein the training data comprises land conditions data correlated with the labeled weather event data. (in at least [0145] Image occlusion is the process to crop out, occlude, or set to zero some portions of a 2D climate data image such as temperature anomaly map 550. For example, a random number of randomly located squares each of a fixed or random size may be cropped out. Occluding part of the climate data image prevents overly focusing on one area of the globe as key for forecasting a climate phenomenon such as El Nino, and tries to extract clues or signals from more geographies than without occlusion. The number, size, shape, and locations of portions of a climate data image to occlude, and the number of occlusion attempts, are hyperparameters of the image occlusion-based data augmentation process. In some embodiments, land-sea masks may be used for occluding selected portions of the climate data image to focus on land only or on sea only, for example when the land is a poor predictor for the desired climate forecast. In an illustrative example, 3-10 augmented climate data images may be generated by occluding random-sized boxes having 10% to 30% of the total area, where data within the random-sized boxes are set to 0. Such an image occlusion process may be run probabilistically, as configured through a hyperparameter, set to a value within a given range, such as between 0.1 and 0.9. The probability of re-running the occlusion process on the same image may also be configured to a value within a given range, such as between 0.1 and 0.3.)
As per Claim 27, Ospina teaches: (Previously Presented) The method of claim 21,
… satellites. (in at least [0061] a GCM is initialized with observed or estimated atmosphere, ocean, land, and sea ice states, and run in forward time into the future. Because of the chaotic nature of the climate system, forecast results can be very sensitive to even small perturbations to the initial conditions or model parameters of the system. Any changes in perturbations or external forcings to the system, for example in the form of solar irradiance, or human contributed carbon and aerosol emissions, would require a GCM-based forecast to be run again, and any additional lead-time for the forecast requires at least polynomial increase in the total number of computations needed, while also increasing the amount of forecast uncertainty. Moreover, differences in GCM model design often lead to very different forecasting skills, with some models performing better than others in some specific climate forecasting applications. Models can also perform better at some specific time of the year than at other times of the year. Ensemble modeling such as seasonal predictions through the North American Multi-Model Ensemble (NMME) has been studied to reduce forecast uncertainty by post-processing, ranking, weighting, and averaging climate projections from different GCMs, yet such approaches may require even higher computational power [0063] machine-learning based climate forecast systems have utilized long short-term memory neural networks or a combination of autoregressive integrated moving average models and artificial neural networks. Such forecasts are trained and validated exclusively on observational historical data, and are thus significantly constrained by the short observational record of climate data, which has only been measured in situ or via satellites on the global scale for the past hundred years or so.)
Although implied, Ospina does not expressly disclose the following limitations, which however, are taught by Tran,
wherein the plurality of sensors comprise satellites. (in at least [0014] a motorized frame to move in a field, cameras, temperature sensors, and other sensors to capture field data, and machine learning to monitor and manage problems in field. One implementation can collect weather data like hail, wind, rainfall, and temperature, which can be placed in fields to assist in data-driven field decisions. One embodiment supplements local weather data with external sources like satellites and radar. This embodiment can correct for what is being provided in the public realm through satellites and radar—making it more precise and accurate—by fusing it with the insight gleaned from irrigation ground-based stations. [0065] to monitor weather, another system includes a motorized frame to move in a field, cameras, temperature sensors, and other sensors to capture field data, and machine learning to monitor and manage problems in field. One implementation can collect weather data like hail, wind, rainfall, and temperature, which can be placed in fields to assist in data-driven field decisions. One embodiment supplements local weather data with external sources like satellites and radar. This embodiment can correct for what is being provided in the public realm through satellites and radar—making it more precise and accurate—by fusing it with the insight gleaned from irrigation ground-based stations. [0189] Other in- or on-ground sensors can be deployed to detect crop conditions, weather data, and many other details, which can then be transmitted to decision analytics platforms via the Internet of Things (where computing devices embedded in everyday objects are connected to the Internet to enable analytics). A solar-powered in-ground sensor can gather data on crop stress, air pressure, humidity, temperature, chlorophyll, canopy biomass, rainfall, and other information, which can then be analyzed on its platform to improve precision farming.)
The reason and rationale to combine Ospina and Tran is the same as recited above.
As per Claim 28, Ospina teaches: (Currently Amended) The method of claim 22, wherein returning the information identifying the particular weather event comprises
publishing, onto a …, a weather event amongst the weather events having the respective significant levels that are greater than the predetermined threshold values. (in at least [0113] the at least one climate variable may be an input climate variable to the forecasting system, one or more climate variables having direct functional dependencies with the input climate variable, or any other set of climate variables having some measurable significance on the target forecasting application. In one instance, the at least one climate variable is sea surface temperature for an El Nino region, and the statistics of interest may include bias, variance, correlation, autocorrelation over time, frequency of peaks, and the like. Sea surface temperature may be chosen here based on the target output variable being the Nino-3.4 index. [0150] Further fine-tuning of the NN may occur at step 940, based on observational historical data 935 including reanalysis data 938. Recall that climate reanalysis combines and assimilates observational historical data with physical dynamical models to “fill in gaps” and provide a physically coherent and consistent, synthesized estimate of the climate in the past, while keeping the historical record uninfluenced by artificial factors. The availability of reanalysis data is limited by that of the observational historical data, and typical reanalysis datasets span over 40 to 80 years. To maximize the effectiveness of reanalysis data 938 in step 940, in some embodiments, some NN layers may be frozen during tuning, with only a selected subset of NN layers further updated. For example, in CRNN 500 shown in FIG. 5A, the first five convolutional layers and the RNN may be frozen, while the last convolutional layer is fine-tuned on 80 years of reanalysis data. [0151] Once trained, the climate forecasting NN may be validated and tested at step 950, using validation and test sets comprising observational historical data. In some embodiments, all available observational historical data including reanalysis data may be divided by year into a tuning set, a validation set, and/or a testing set, for use in steps 940 and 950. Based on validation results that indicate forecast uncertainty or confidence, any of the steps 910, 920, 930 and 940 may be repeated to update the NN climate forecasting model and improve performance of the forecasting system. The fully trained deployed in step 950 for actual climate forecasting. [0065] the CLIMATEAI climate forecasting system employs a deep learning network that is capable of extracting spatial-temporal features as well as functional dependencies and correlations among different GCM simulation datasets to predict future climate conditions. Typically in supervised learning, a predictor model such as a neural network is first trained using a first set of labeled training data to determine an optimal set of internal parameters. The capability of the predictor model is then validated on a second validation dataset and tuned accordingly. A third test dataset is then used to evaluate the predictive or forecast skill of the model. [0115] At step 740, validated GCMs or GCM simulation datasets are further evaluated for their ability to forecast the target output climate variable, or for their ability to forecast some climate variables highly correlated with the target output climate variable. In some embodiments, a forecast skill score is computed for each validated GCM based on a forecast function, where the forecast function may be a data predictor function, the NN-based climate forecasting model which will be trained using data ensemble 790, or a model-analog. [0164] FIG. 13A is a graph 1300 comparing correlation measures of different forecasting methods in predicting the Nino 3.4 Index at different lead times, according to some embodiments of the present invention. Meanwhile, FIG. 13B is a graph 1350 comparing the time series of forecast results by different forecasting methods in predicting the Nino 3.4 Index, according to some embodiments of the present invention. The CLIMATEAI forecast result shown in FIGS. 13A and 13B are based on an NN trained on a data ensemble that interleaves simulation data from multiple GCMs. It can be seen that neural networks trained on AOGCMs according to embodiments of the present invention offer comparable forecasting performance to SEAS5 while consuming only a miniscule fraction of the computation power. The CLIMATEAI system that learns from abundant AOGCM simulations also outperforms NNs trained purely on limited amount of historical observations. [0169] FIG. 16 is an illustrative diagram 1600 showing an analysis of power generation by a hydroelectric plant including seasonal variations in power production, according to some embodiments of the present invention. Total generations and river stream flows are compared in the bottom panel, indicating a high correlation with seasonality in power generation at this site. That is, FIG. 15 illustrates yet another GUI that can be accessed by a user to access or operate a system implementing some embodiments of the present invention. [0172] the user selection includes an industry and one or more parameters associated with the industry. In some embodiments, the industry is an energy industry, a real estate industry, an agricultural industry, a financial industry, and/or an insurance industry. In some embodiments, the energy industry includes a wind power industry, and the one or more parameters include one or more of a location, a wind power farm, and a forecast horizon. In some embodiments, the energy industry includes a hydroelectric industry, and the one or more parameters include one or more of a location, a hydroelectric dam, and a forecast horizon.)
Although implied, Ospina does not expressly disclose the following limitations, which however, are taught by Tran,
… onto a ledger… (in at least [0005] The seed comprises hemp and the blockchain comprises Hyperledger or Ethereum blockchain. The end product properties include product size, drought resistance, or predetermined taste. The system can add entries to the blockchain as the seed moves from the farm to a consumer. The system can dispense fertilizer in response to the growth data. The microbes can be embedded in a fertilizer. The seed can be coated with the microbes. Data can be received from IoT (internet of things) sensors and communicating with a cloud to control a farm system. A lending source (bank/government agency) can finance the seed based on blockchain data. The system can use cryptocurrency by generating tokens used by farmers, farm suppliers, shippers, distributors, and retailers in an ecosystem. Each generation of the microbes are iteratively grown in a feedback loop. Each generation of the microbes are iteratively grown to increase microbes providing the predetermined end product property, comprising checking genetic data of each generation of microbes in a feedback loop to increase or decrease predetermined members of the microbes to achieve the determined end product property. Sensors can detect humidity, temperature, leaf property, soil property, or presence of an organism on the plant. A camera can capture a leaf image and determining soil properties based from the leaf image. The camera can capture a leaf image and determining soil properties based from the leaf image. The camera can capture soil image and determining soil properties based therefrom. The camera can capture an image of an organism on the plant, classifying the organism, and recommending a treatment for the organism. The system can communicate over 5G cellular network. A farmer can get the seed and a third party can agree to buy an end product from the seed with the predetermined end product properties in advance of planting. A network of trucks can pick-up the end product by selecting in real-time one or more nearest trucks from a network of trucks, reviewing each truck's shipping cost, and designating one of the trucks to transport the end product. [0364] The complete history and current location of any food item along with its accompanying information (i.e. certifications, test data, temperature data) can be readily available in seconds. The system combines supply chain modules with blockchain core functions, delivering business value to the food ecosystem from the combination of governance, standards and interoperability, and technology. All data is stored on blockchain ledgers, protected with the highest level of commercially-available, tamper-resistant encryption. The solution provides participants with a permission-based, shared view of food ecosystem information, allowing convenient data publishing and controlled sharing of information.)
The reason and rationale to combine Ospina and Tran is the same as recited above.
As per Claim 29, Ospina teaches: (Currently Amended) The method of claim 22, wherein returning the information identifying the particular weather event comprises
publishing, in a …, a weather event amongst the weather events having the respective significant levels that are greater than the predetermined threshold values. (in at least [0113] the at least one climate variable may be an input climate variable to the forecasting system, one or more climate variables having direct functional dependencies with the input climate variable, or any other set of climate variables having some measurable significance on the target forecasting application. In one instance, the at least one climate variable is sea surface temperature for an El Nino region, and the statistics of interest may include bias, variance, correlation, autocorrelation over time, frequency of peaks, and the like. Sea surface temperature may be chosen here based on the target output variable being the Nino-3.4 index. [0150] Further fine-tuning of the NN may occur at step 940, based on observational historical data 935 including reanalysis data 938. Recall that climate reanalysis combines and assimilates observational historical data with physical dynamical models to “fill in gaps” and provide a physically coherent and consistent, synthesized estimate of the climate in the past, while keeping the historical record uninfluenced by artificial factors. The availability of reanalysis data is limited by that of the observational historical data, and typical reanalysis datasets span over 40 to 80 years. To maximize the effectiveness of reanalysis data 938 in step 940, in some embodiments, some NN layers may be frozen during tuning, with only a selected subset of NN layers further updated. For example, in CRNN 500 shown in FIG. 5A, the first five convolutional layers and the RNN may be frozen, while the last convolutional layer is fine-tuned on 80 years of reanalysis data. [0151] Once trained, the climate forecasting NN may be validated and tested at step 950, using validation and test sets comprising observational historical data. In some embodiments, all available observational historical data including reanalysis data may be divided by year into a tuning set, a validation set, and/or a testing set, for use in steps 940 and 950. Based on validation results that indicate forecast uncertainty or confidence, any of the steps 910, 920, 930 and 940 may be repeated to update the NN climate forecasting model and improve performance of the forecasting system. The fully trained deployed in step 950 for actual climate forecasting. [0065] the CLIMATEAI climate forecasting system employs a deep learning network that is capable of extracting spatial-temporal features as well as functional dependencies and correlations among different GCM simulation datasets to predict future climate conditions. Typically in supervised learning, a predictor model such as a neural network is first trained using a first set of labeled training data to determine an optimal set of internal parameters. The capability of the predictor model is then validated on a second validation dataset and tuned accordingly. A third test dataset is then used to evaluate the predictive or forecast skill of the model. [0115] At step 740, validated GCMs or GCM simulation datasets are further evaluated for their ability to forecast the target output climate variable, or for their ability to forecast some climate variables highly correlated with the target output climate variable. In some embodiments, a forecast skill score is computed for each validated GCM based on a forecast function, where the forecast function may be a data predictor function, the NN-based climate forecasting model which will be trained using data ensemble 790, or a model-analog. [0164] FIG. 13A is a graph 1300 comparing correlation measures of different forecasting methods in predicting the Nino 3.4 Index at different lead times, according to some embodiments of the present invention. Meanwhile, FIG. 13B is a graph 1350 comparing the time series of forecast results by different forecasting methods in predicting the Nino 3.4 Index, according to some embodiments of the present invention. The CLIMATEAI forecast result shown in FIGS. 13A and 13B are based on an NN trained on a data ensemble that interleaves simulation data from multiple GCMs. It can be seen that neural networks trained on AOGCMs according to embodiments of the present invention offer comparable forecasting performance to SEAS5 while consuming only a miniscule fraction of the computation power. The CLIMATEAI system that learns from abundant AOGCM simulations also outperforms NNs trained purely on limited amount of historical observations. [0169] FIG. 16 is an illustrative diagram 1600 showing an analysis of power generation by a hydroelectric plant including seasonal variations in power production, according to some embodiments of the present invention. Total generations and river stream flows are compared in the bottom panel, indicating a high correlation with seasonality in power generation at this site. That is, FIG. 15 illustrates yet another GUI that can be accessed by a user to access or operate a system implementing some embodiments of the present invention. [0172] the user selection includes an industry and one or more parameters associated with the industry. In some embodiments, the industry is an energy industry, a real estate industry, an agricultural industry, a financial industry, and/or an insurance industry. In some embodiments, the energy industry includes a wind power industry, and the one or more parameters include one or more of a location, a wind power farm, and a forecast horizon. In some embodiments, the energy industry includes a hydroelectric industry, and the one or more parameters include one or more of a location, a hydroelectric dam, and a forecast horizon.)
Although implied, Ospina does not expressly disclose the following limitations, which however, are taught by Tran,
…in a transaction record… (in at least [0005] The seed comprises hemp and the blockchain comprises Hyperledger or Ethereum blockchain. The end product properties include product size, drought resistance, or predetermined taste. The system can add entries to the blockchain as the seed moves from the farm to a consumer. The system can dispense fertilizer in response to the growth data. The microbes can be embedded in a fertilizer. The seed can be coated with the microbes. Data can be received from IoT (internet of things) sensors and communicating with a cloud to control a farm system. A lending source (bank/government agency) can finance the seed based on blockchain data. The system can use cryptocurrency by generating tokens used by farmers, farm suppliers, shippers, distributors, and retailers in an ecosystem. Each generation of the microbes are iteratively grown in a feedback loop. Each generation of the microbes are iteratively grown to increase microbes providing the predetermined end product property, comprising checking genetic data of each generation of microbes in a feedback loop to increase or decrease predetermined members of the microbes to achieve the determined end product property. Sensors can detect humidity, temperature, leaf property, soil property, or presence of an organism on the plant. A camera can capture a leaf image and determining soil properties based from the leaf image. The camera can capture a leaf image and determining soil properties based from the leaf image. The camera can capture soil image and determining soil properties based therefrom. The camera can capture an image of an organism on the plant, classifying the organism, and recommending a treatment for the organism. The system can communicate over 5G cellular network. A farmer can get the seed and a third party can agree to buy an end product from the seed with the predetermined end product properties in advance of planting. A network of trucks can pick-up the end product by selecting in real-time one or more nearest trucks from a network of trucks, reviewing each truck's shipping cost, and designating one of the trucks to transport the end product. [0364] The complete history and current location of any food item along with its accompanying information (i.e. certifications, test data, temperature data) can be readily available in seconds. The system combines supply chain modules with blockchain core functions, delivering business value to the food ecosystem from the combination of governance, standards and interoperability, and technology. All data is stored on blockchain ledgers, protected with the highest level of commercially-available, tamper-resistant encryption. The solution provides participants with a permission-based, shared view of food ecosystem information, allowing convenient data publishing and controlled sharing of information.)
The reason and rationale to combine Ospina and Tran is the same as recited above.
As per Claim 30, Ospina teaches: (Previously Presented) The method of claim 21, wherein the method further comprises:
receiving historic data associated with the weather event data; and (in at least [0152] FIG. 10 is another exemplary flow diagram 1000 for a process to generate and train an NN-based climate forecasting model, according to some embodiments of the present invention. Upon initialization at step 1005, global climate simulation data 1006 from at least two global climate simulation models are combined at step 1010 into a multi-model ensemble. The multi-model ensemble is then pre-processed at step 1020, where the pre-processing comprises at least one pre-processing action selected form the group consisting of spatial re-gridding, temporal homogenization, and data augmentation. At step 1030, a neural network (NN)-based climate forecasting model is trained on the pre-processed multi-model global climate simulation data ensemble, and validated on a set of observational historical climate data at step 1040. The observational historical climate data may comprise reanalysis data, and the NN's hyperparameters may be fine-tuned during the validation process as well. At an optional step 1050, the NN-based climate forecasting model may be deployed, and the process ends at step 1060.)
providing (i) the weather event data and (ii) the historic data associated with the weather event data as inputs to the trained neural network, wherein the neural network was trained to detect hidden correlations between the weather event data that occur in earlier periods of time and events in the weather event data that occur in later periods of time. (in at least [0067] The second novel feature of the CLIMATEAI system is its ability to pre-process the multi-model data ensemble to reduce or remove data heterogeneity, and to augment the data ensemble further, reinforcing the underlying hidden functional dependencies among different simulated climate datasets. [0101] NN 500 is a Convolutional Recurrent Neural Network (CRNN) with a convolutional and recurrent architecture: it encodes the spatial information of each global surface temperature grid using a Convolutional Neural Network (CNN) first, then feeds the encoded information into a Recurrent Neural Network (RNN) having Long Short-Term Memory (LSTM) layers to learn from the temporal sequence. A CNN utilizes the process of convolution to reduce the number of model parameters, and to capture the spatial dependencies in input data. An RNN, on the other hand, has connections that form a directed graph along a temporal sequence, to recognize sequential characteristics and patterns within the input data to predict a future event or scenario. [0102] NN 500 first feeds 2-dimensional inputs 510 through multiple convolution (Conv2D) layers with Rectified Linear Units (ReLU), then a fully connected (FC) layer 520. In this illustrative embodiment, the CNN comprises 6 layers with the following network details: [Conv2D=>batch normalization=>ReLU]×5=>[Conv2D=>ReLU]=>FC. A convolutional layer applies a convolution or correlation operation by a kernel matrix to the input data to generate a feature map of the input image. ReLU is a non-linear activation function. A fully connected layer has full connections to all activations in the previous layer, and is needed before classification or output activation at an output layer of the NN. Successive convolution-ReLU-pooling stages allow the successive extraction of low-level to high-level features, from local temperature correlations to distant teleconnections. The first convolutional layer in FIG. 5A may use 10 filters, and the number of filters may double in every subsequent convolutional layer. Paddings and strides may be defined to get desired size reductions. Output vector 525 from fully connect layer 520 may feed into an RNN 530 in sequences of successive months, such as 24 months. RNN 530 may have a many-to-one architecture, and may use two LSTM layers, each having 500 hidden units. At the end, the hidden state of the last time step may be decoded to a real value using another fully connected layer, to output predicted monthly Nino-3.4 sea surface temperature anomalies. While the many-to-one architecture may be used in some embodiments, in other embodiments, a many-to-many architecture may also be used in forecasting multiple months simultaneously. [0152] FIG. 10 is another exemplary flow diagram 1000 for a process to generate and train an NN-based climate forecasting model, according to some embodiments of the present invention. Upon initialization at step 1005, global climate simulation data 1006 from at least two global climate simulation models are combined at step 1010 into a multi-model ensemble. The multi-model ensemble is then pre-processed at step 1020, where the pre-processing comprises at least one pre-processing action selected form the group consisting of spatial re-gridding, temporal homogenization, and data augmentation. At step 1030, a neural network (NN)-based climate forecasting model is trained on the pre-processed multi-model global climate simulation data ensemble, and validated on a set of observational historical climate data at step 1040. The observational historical climate data may comprise reanalysis data, and the NN's hyperparameters may be fine-tuned during the validation process as well. At an optional step 1050, the NN-based climate forecasting model may be deployed, and the process ends at step 1060.)
As per Claim 31, Ospina teaches: (Previously Presented) The method of claim 21, wherein the plurality of … are further configured to
generate weather forecast data, the weather forecast data being provided as additional inputs to the trained neural network. (in at least [0106] The training process begins at step 610 with data acquisition, retrieval, assimilation, or generation. At step 620, acquired data are pre-processed, or prepared. At step 630, the ML model is trained using training data 625. At step 640, the ML model is evaluated, validated, and tested, and further refinements to the ML model are fed back into step 630 for additional training. Once its performance is acceptable, at step 650, optimal model parameters are selected, for deployment at step 660. New data 655 may be used by the deployed model to make predictions. [0135] MDA may also extract a global forcing signal (e.g., global warming), and an internal variability signal (e.g., weather). MDA first uses linear regression to learn the influences of ENSO, AMO, and PDO on a climate variable at each location, then represents the value of the climate variable as the sum of six components: the mean value of the variable, the influences of ENSO, the influence of AMO, the influence of PDO, the influence of climate change, and internal climate variability. MDA further uses surrogate time series generation techniques to generate alternate ENSO, AMO and PDO time series with means, variances, and autocorrelations similar to the original time series, and bootstraps to generate alternate estimates of internal climate variability. MDA then substitutes in new values of the influences of ENSO, AMO, and PDO, and the effect of internal climate variability into its representation of the climate system as the sum of different components. This process is repeated for each location on earth multiple times to generate alternate viable climate states, which may in turn be used for GCM data augmentation. [0152] FIG. 10 is another exemplary flow diagram 1000 for a process to generate and train an NN-based climate forecasting model, according to some embodiments of the present invention. Upon initialization at step 1005, global climate simulation data 1006 from at least two global climate simulation models are combined at step 1010 into a multi-model ensemble. The multi-model ensemble is then pre-processed at step 1020, where the pre-processing comprises at least one pre-processing action selected form the group consisting of spatial re-gridding, temporal homogenization, and data augmentation. At step 1030, a neural network (NN)-based climate forecasting model is trained on the pre-processed multi-model global climate simulation data ensemble, and validated on a set of observational historical climate data at step 1040. The observational historical climate data may comprise reanalysis data, and the NN's hyperparameters may be fine-tuned during the validation process as well. At an optional step 1050, the NN-based climate forecasting model may be deployed, and the process ends at step 1060.)
Although implied, Ospina does not expressly disclose the following limitations, which however, are taught by Tran,
…plurality of sensors… (in at least [0014] a motorized frame to move in a field, cameras, temperature sensors, and other sensors to capture field data, and machine learning to monitor and manage problems in field. One implementation can collect weather data like hail, wind, rainfall, and temperature, which can be placed in fields to assist in data-driven field decisions. One embodiment supplements local weather data with external sources like satellites and radar. This embodiment can correct for what is being provided in the public realm through satellites and radar—making it more precise and accurate—by fusing it with the insight gleaned from irrigation ground-based stations. [0065] to monitor weather, another system includes a motorized frame to move in a field, cameras, temperature sensors, and other sensors to capture field data, and machine learning to monitor and manage problems in field. One implementation can collect weather data like hail, wind, rainfall, and temperature, which can be placed in fields to assist in data-driven field decisions. One embodiment supplements local weather data with external sources like satellites and radar. This embodiment can correct for what is being provided in the public realm through satellites and radar—making it more precise and accurate—by fusing it with the insight gleaned from irrigation ground-based stations. [0189] Other in- or on-ground sensors can be deployed to detect crop conditions, weather data, and many other details, which can then be transmitted to decision analytics platforms via the Internet of Things (where computing devices embedded in everyday objects are connected to the Internet to enable analytics). A solar-powered in-ground sensor can gather data on crop stress, air pressure, humidity, temperature, chlorophyll, canopy biomass, rainfall, and other information, which can then be analyzed on its platform to improve precision farming.)
The reason and rationale to combine Ospina and Tran is the same as recited above.
As per Claim 38, Ospina teaches: (Previously Presented) The method of claim 32,
wherein the machine learning model is a Long Short-Term Memory (LSTM) model. (in at least [0101] NN 500 is a Convolutional Recurrent Neural Network (CRNN) with a convolutional and recurrent architecture: it encodes the spatial information of each global surface temperature grid using a Convolutional Neural Network (CNN) first, then feeds the encoded information into a Recurrent Neural Network (RNN) having Long Short-Term Memory (LSTM) layers to learn from the temporal sequence. A CNN utilizes the process of convolution to reduce the number of model parameters, and to capture the spatial dependencies in input data. An RNN, on the other hand, has connections that form a directed graph along a temporal sequence, to recognize sequential characteristics and patterns within the input data to predict a future event or scenario.)
As per Claim 32-37 for a method (see at least Ospina [0069]), substantially recite the subject matter of Claim 21 and are rejected based on the same reasoning and rationale.
As per Claim 39-40 for a method (see at least Ospina [0069]), substantially recite the subject matter of Claim 21 and are rejected based on the same reasoning and rationale.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any extension fee pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to PO HAN MAX LEE whose telephone number is (571)272-3821. The examiner can normally be reached on Mon-Thurs 8:00 am - 7:00 pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Rutao Wu can be reached on (571) 272-6045. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/PO HAN LEE/Primary Examiner, Art Unit 3623
Prosecution Insights