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
This office action is in response to the claims filed on 11/05/2021.
Claims1-20 are presented for examination.
Information Disclosure Statement
The information disclosure statement filed 04/11/2024 fails to comply with 37 CFR 1.98(a)(2), which requires a legible copy of each cited foreign patent document; each non-patent literature publication or that portion which caused it to be listed; and all other information or that portion which caused it to be listed. It has been placed in the application file, but the information referred to therein has not been considered. .
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 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 analysis:
In the instant case, the claims are directed to a method (claims 1-10), system (claims 11, 12-20). Thus, each of the claims falls within one of the four statutory categories (i.e., process, machine, manufacture, or composition of matter).
Step 2A analysis:
Based on the claims being determined to be within of the four categories (Step 1), it must be determined if the claims are directed to a judicial exception (i.e., law of nature, natural phenomenon, and abstract idea), in this case the claims fall within the judicial exception of an abstract idea. Specifically, the abstract idea of “Mental Processes/Concepts performed in the human mind (including an observation, evaluation, judgment, opinion)” and mathematical concept.
The claim 1 recites:
a) Step 2A: prong 1 analysis:
-“ down sampling a time series dataset to generate an initial input having a first scale resolution, such that the first scale resolution is less than a scale resolution of the time series dataset” this is a mental process, as the human mind can down sampling the data based on the pattern or on the particular purpose of using the data, for example, the human may reduce some data point that is not relevant to the current use purpose, (observation/evaluation)
upsampling by an upsampling function the first output to generate a second input having a second scale resolution, the second scale resolution being higher than the first scale resolution, such that the second input is based on the first output; this is a mental process, as the human mind can up sampling the data based on the pattern or on the particular purpose of using the data, for example, the human may add some more data point that is relevant to the current use purpose, (observation/evaluation),
“processing as a first iteration, the initial input to generate a first output, processing as a second iteration, the second input to generate a second output; wherein the second output represents a time series forecast of the time series dataset.” It’s a mental process, the human mind can iteratively receive the input (first and second received some information) and generating the output (first and second output), (observation/Evaluation). “.
Step 2A: Prong 2 analysis:
-“ operating a neural network using an encoder-based model to provide a time series forecast,”, “processing as a first iteration, using the model, the initial input”, “processing as a second iteration, using the model, the second input” The additional limitation is recited at high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer component (encoder, model) (See MPEP 2106.05(f)).
b) Step 2B analysis:
-“ operating a neural network using an encoder-based model to provide a time series forecast,”, “ operating a neural network using an encoder-based model to provide a time series forecast,”, “processing as a first iteration, using the model, the initial input”, “processing as a second iteration, using the model, the second input” The additional limitation is recited at high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer component (encoder, model) (See MPEP 2106.05(f)).
The claim 2 recites:
a) Step 2A: prong 1 analysis:
-“ comprising continuing to iterate using one or more subsequent iterations and the upsampling function until a resolution scale of the time series forecast matches the scale resolution of the time series dataset.” This is a mental process, the human can iteratively apply the up sampling function until the resolution scale of the time series forecast match the scale resolution of the time series dataset, for example, the person can re-scale the time series forecast that match the particular scale resolution without the computer, (Observation/Evaluation).
Step 2A: Prong 2 analysis:
- “continuing to iterate using one or more subsequent iterations using the model” The additional element is recited at high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer component (See MPEP 2106.05(f)).
b) Step 2B analysis:
- “continuing to iterate using one or more subsequent iterations using the model” The additional element is recited at high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer component (See MPEP 2106.05(f)).
The claim 3 recites:
a) Step 2A: prong 1 analysis:
-“ wherein a resolution scale of the time series forecast matches the scale resolution of the time series dataset.” This is a mental process, as the human mind can match the resolution scale of the time series forecast to the scale resolution of the time series dataset, for example, the person can re-scale the time series forecast that match the particular scale resolution without the computer (observation/Evaluation).
Step 2A: Prong 2 analysis and Step 2B analysis
No additional element that provides a practical application or amount to significantly more than the abstract idea.
The claim 4 recites:
a) Step 2A: prong 2 analysis:
-“ using a same encoder for each of the first iteration and the second iteration.” The additional element is recited at high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer component (See MPEP 2106.05(f)).
b) Step 2B analysis:
-“ using a same encoder for each of the first iteration and the second iteration.” The additional element is recited at high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer component (See MPEP 2106.05(f)).
The claim 5 recites:
Step 2A: Prong 2 analysis:
-“ using a different encoder for each of the first iteration and the second iteration.” The additional element is recited at high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer component (See MPEP 2106.05(f)).
b) Step 2B analysis:
-“ using a different encoder for each of the first iteration and the second iteration.” The additional element is recited at high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer component (See MPEP 2106.05(f)).
The claim 6 recites:
a) Step 2A: prong 1 analysis:
-“ using a normalization function on the initial input in order to normalize the initial input before said processing using the model.” This is a mathematical concept.
Step 2A: Prong 2 analysis and Step 2B analysis
No additional element that provides a practical application or amount to significantly more than the abstract idea.
The claim 7 recites:
a) Step 2A: prong 1 analysis:
-“ using a normalization function on the second input in order to normalize the second input before said processing using the model.” This is a mathematical concept.
Step 2A: Prong 2 analysis and Step 2B analysis
No additional element that provides a practical application or amount to significantly more than the abstract idea.
The claim 8 recites:
a) Step 2A: prong 1 analysis:
-“ using a loss function on the second output in order to quantify a error present in the time series forecast.” This is a mathematical concept.
Step 2A: Prong 2 analysis and Step 2B analysis
No additional element that provides a practical application or amount to significantly more than the abstract idea.
The claim 9 recites:
a) Step 2A: prong 2 analysis:
-“ the model is a transformer model.” The additional element is recited at high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer component (See MPEP 2106.05(f)).
b) Step 2B analysis:
-“ the model is a transformer model.” The additional element is recited at high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer component (See MPEP 2106.05(f)).
The claim 10 recites:
a) Step 2A: prong 2 analysis:
-“ the model is a probabilistic model.” The additional element is recited at high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer component (See MPEP 2106.05(f)).
b) Step 2B analysis:
-“ the model is a probabilistic model.” The additional element is recited at high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer component (See MPEP 2106.05(f)).
The claim 11 recites:
a) Step 2A: prong 2 analysis:
-“ An artificial neural network operated in accordance with the method of claim 1” The additional element is recited at high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer component (See MPEP 2106.05(f)).
b) Step 2B analysis:
-“ An artificial neural network operated in accordance with the method of claim 1” The additional element is recited at high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer component (See MPEP 2106.05(f)).
The claim 12 recites:
a) Step 2A: prong 2 analysis:
-“a database storing a time series dataset that is communicatively coupled to the processor; and a memory that is communicatively coupled to the processor and that has stored thereon computer program code that is executable by the processor and that, when executed by the processor,” The additional element is recited at high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer component (See MPEP 2106.05(f)).
“causes the processor to retrieve the time series dataset from the database and to use the time series dataset to perform the method of claim 1” .These/this additional limitation(s) are/is recited at a high-level of generality such that it amounts to necessary data gathering. As described in MPEP 2106.05(g), limitations that amount to merely adding insignificant extra-solution activity of data gathering to a judicial exception do not amount to significantly more than the judicial exception and cannot integrate a judicial exception into a practical application.
b) Step 2B analysis:
--“a database storing a time series dataset that is communicatively coupled to the processor; and a memory that is communicatively coupled to the processor and that has stored thereon computer program code that is executable by the processor and that, when executed by the processor,” The additional element is recited at high level of generality and amounts to no more than mere instructions to apply the judicial exception using a generic computer component (See MPEP 2106.05(f)).
“causes the processor to retrieve the time series dataset from the database and to use the time series dataset to perform the method of claim 1.” These/this limitation(s) are/is recited at a high-level of generality such that it amounts to necessary data gathering. As described in MPEP 2106.05(g), limitations that amount to merely adding insignificant extra-solution activity of data gathering to a judicial exception do not amount to significantly more than the judicial exception itself .
The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory"). (this evidence is applied for data gathering/storing data).
The claims 13-20 are rejected for the same reason as the claims 2-9, since these claims recite the same limitations.
Additionally, the claim 11 is rejected under 35 U.S.C. 101 because the claimed invention is directed to non-statutory subject matter. The claim is directed to a program, therefore, this is software per se. A claim that recites a piece of software alone without any link to a hardware component is directed to non-statutory subject matter since there is no relationship between the computer software and hardware components which permits the functionality of the software to be realized. The claims lack the necessary physical articles or objects to constitute a machine or a manufacture within the meaning of 35 USC 101. The claim recites that the program, when executed by a computer, causes a certain function, however, the claim is directed to the program by itself, not to a device that comprises a processor that executes a program, or a computer program product. As such, it fails to fall within a statutory category under 35 USC 101.
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 for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1, 10, 11, 12 are rejected under 35 U.S.C. 103 as being unpatentable over KO et al. (PUB. No 20220187486-hereinafter, KO) in view of Viswanathan et al. (PUB. No 20180293706 -hereinafter, Viswanathan).
Regarding claim 1, KO teaches a method for operating a neural network using an encoder-based model to provide a time series forecast, the method comprising (KO, [Par.0009, Fig.4A, 4B], “The computer system comprises at least one computer including an arithmetic unit and a storage device coupled to the arithmetic unit, and manages model information for defining a U-Net configured to execute, on the input time-series data, an encoding operation for extracting a feature map relating to the target wave through use of a plurality of downsampling blocks and a decoding operation for outputting data for predicting the first motion time of the target wave through use of a plurality of upsampling blocks. The at least one computer is configured to execute the encoding operation and the decoding operation on the input time-series data through use of the model information.. …” and [Par.0060-0061], “The downsampling block 300 includes a one-dimensional convolutional layer 400, two one-dimensional residual blocks 401, and a one-dimensional max pooling layer 402. The structure of the downsampling block 300 illustrated in FIG. 4A is merely an example, and is not limited thereto. Some of the components may be excluded, or another component may be included. In addition, the order of input and output may be changed.[0061] The upsampling block 310 includes a one-dimensional upsampling layer 403, a connected layer 404, a one-dimensional convolutional layer 400, and two one-dimensional residual blocks 401. The structure of the upsampling block 310 illustrated in FIG. 4B is merely an example, and is not limited thereto. Some of the components may be excluded, or another component may be included. In addition, the order of input and output may be changed.” Examiner’s note, the computer is configured to execute the encoding operation, therefore, the computer is considered as the encoder.):
down sampling a time series dataset to generate an initial input having a first scale resolution, such that the first scale resolution is less than a scale resolution of the time series dataset (KO, [Par. 0051-0055] “FIG. 2 is a diagram for illustrating a structure of a model in the first embodiment. FIG. 3A is a graph for showing an example of time-series data on a wave input to the model in the first embodiment. FIG. 3B is a graph for showing an example of data output from the model in the first embodiment. FIG. 3C is a graph for showing an example of teacher data in the first embodiment. [0052] The model in the first embodiment is a model based on a U-Net described in Olaf Ronneberger and two others, “U-Net: Convolutional Networks for Biomedical Image Segmentation,” retrieved on Nov. 2, 2020 through the Internet. The model in the first embodiment includes four tiers of downsampling blocks 300 for implementing an encoding operation for extracting a feature, and four tiers of upsampling blocks 310 for implementing a decoding operation. [0055], In FIG. 2, the numbers by the solid arrows each represent numbers of time steps and channels of input or output data. For example, the time-series data on the wave having a time step of 10,000 is input to the downsampling block 300 in the first tier, and eight feature maps having a time step of 2,000 (the number of channels being 8) are output.” Examiner’s note, the time series data is input into the model, wherein, the model include the four tier of the downsampling block and four tier of upsampling block, therefore, the generating of the down-sampling blocks on the time serries data. the down sampling the time-serries data (time step of 10,000) input into the down sampling block to generate the output with the time step of 2,000 (the first scale resolution), therefore, the first scale resolution is 2000 is less than the scale resolution of the time series data 10,000.);
processing as a first iteration, using the model, the initial input to generate a first output (KO, [Par 0051-0055, Fig. 2, 3A, 3B], “FIG. 2 is a diagram for illustrating a structure of a model in the first embodiment. FIG. 3A is a graph for showing an example of time-series data on a wave input to the model in the first embodiment. FIG. 3B is a graph for showing an example of data output from the model in the first embodiment. FIG. 3C is a graph for showing an example of teacher data in the first embodiment...[0052], The model in the first embodiment includes four tiers of downsampling blocks 300 for implementing an encoding operation for extracting a feature, and four tiers of upsampling blocks 310 for implementing a decoding operation. [0055] In FIG. 2, the numbers by the solid arrows each represent numbers of time steps and channels of input or output data. For example, the time-series data on the wave having a time step of 10,000 is input to the downsampling block 300 in the first tier, and eight feature maps having a time step of 2,000 (the number of channels being 8) are output.” Examiner’s note, the time series data is input into the model, wherein, the model include the four tier of the downsampling block and four tier of upsampling block, the Fig2, shows processing the down sampling processes and the processing of the up sampling blocks, therefore, the generating of the down-sampling blocks is considered as the first iteration to generating the first output.);
upsampling by an upsampling function the first output to generate a second input having a second scale resolution (KO, [Par.0051-0060, “[0051], “FIG. 2 is a diagram for illustrating a structure of a model in the first embodiment. FIG. 3A is a graph for showing an example of time-series data on a wave input to the model in the first embodiment. FIG. 3B is a graph for showing an example of data output from the model in the first embodiment. FIG. 3C is a graph for showing an example of teacher data in the first embodiment…[0061], The upsampling block 310 includes a one-dimensional upsampling layer 403, a connected layer 404, a one-dimensional convolutional layer 400, and two one-dimensional residual blocks 401. The structure of the upsampling block 310 illustrated in FIG. 4B is merely an example, and is not limited thereto. Some of the components may be excluded, or another component may be included. In addition, the order of input and output may be changed. ” Examiner’s note, the time series data is input into the model, wherein, the model include the four tier of the downsampling block and four tier of upsampling block, the fig.2 shows the output from the down-sampling block is inputting into the up-sampling blocks, therefore, up-sampling block generates input (second input) based on the first output from the down sampling block, that is corresponding to up-sampling by an up-sampling function the first output to generate a second input having a second scale resolution.),
such that the second input is based on the first output (KO,[Par.0051-0061], “[0051], “FIG. 2 is a diagram for illustrating a structure of a model in the first embodiment. FIG. 3A is a graph for showing an example of time-series data on a wave input to the model in the first embodiment. FIG. 3B is a graph for showing an example of data output from the model in the first embodiment. FIG. 3C is a graph for showing an example of teacher data in the first embodiment…[0055]In FIG. 2, the numbers by the solid arrows each represent numbers of time steps and channels of input or output data. For example, the time-series data on the wave having a time step of 10,000 is input to the downsampling block 300 in the first tier, and eight feature maps having a time step of 2,000 (the number of channels being 8) are output. The dotted arrows from the downsampling blocks 300 to the upsampling blocks 310 each indicate connection.. “Examiner’s note, the fig.2 shows the output from the down-sampling block is inputting into the up-sampling blocks, therefore, the second input is based on the first output,);
and processing as a second iteration, using the model, the second input to generate a second output; (KO,[Par.0051-0061], “[0051], “FIG. 2 is a diagram for illustrating a structure of a model in the first embodiment. FIG. 3A is a graph for showing an example of time-series data on a wave input to the model in the first embodiment. FIG. 3B is a graph for showing an example of data output from the model in the first embodiment. FIG. 3C is a graph for showing an example of teacher data in the first embodiment…[0055]In FIG. 2, the numbers by the solid arrows each represent numbers of time steps and channels of input or output data. For example, the time-series data on the wave having a time step of 10,000 is input to the downsampling block 300 in the first tier, and eight feature maps having a time step of 2,000 (the number of channels being 8) are output. The dotted arrows from the downsampling blocks 300 to the upsampling blocks 310 each indicate connection..” Examiner’s note, the fig.2 shows the output from the down-sampling block is inputting into the up-sampling blocks, therefore, the process of the up-sampling is considered as the second iteration to generate the second output based on the second input (output from the down-sampling), .).
wherein the second output represents a time series forecast of the time series dataset (Ko, [Par.0009], “A representative example of the present invention disclosed in this specification is as follows: a computer system for receiving time-series data as input and predicting a first motion time of a target wave. The computer system comprises at least one computer including an arithmetic unit and a storage device coupled to the arithmetic unit, and manages model information for defining a U-Net configured to execute, on the input time-series data, an encoding operation for extracting a feature map relating to the target wave through use of a plurality of downsampling blocks and a decoding operation for outputting data for predicting the first motion time of the target wave through use of a plurality of upsampling blocks.”.)
However, Ko does not teach “the second scale resolution being higher than the first scale resolution”,
On the other hand, Viswanathan the second scale resolution being higher than the first scale resolution ((Viswanathan, [par.0052], “In an example embodiment, the scale of a low resolution image may refer to the up-sampling factor required to generate a higher resolution image of the pre-defined and/or configurable resolution and/or size based on the low resolution image. For example, the up-sampling network may be trained to generate higher resolution images of a pre-defined and/or configurable resolution and/or size independent of the starting resolution and/or size of the low resolution image. For example, the up-sampling network may be trained to receive a first low resolution image at a first scale and use a first set of network weights and/or parameters to generate a first higher resolution image of the pre-defined and/or configurable resolution and/or size and to receive a second low resolution image at a second scale and use a second set of network weights and/or parameters to generate a second higher resolution image of the pre-defined and/or configurable resolution and/or size. The up-sampling network may be trained to receive a third low resolution image at a third scale that is between the first scale and the second scale and to generate a third higher resolution image of the pre-defined and/or configurable resolution and/or size using the a third set of network weights and/or parameters that is determined by (e.g., non-linearly) interpolating between the first and second set of network weights and/or parameters. An interpolation function and the corresponding weights and/or parameters may be determined and/or learned based on the loss function and/or error-weight relationship at the different scales..”.),
Ko and Viswanathan are analogous in arts because they have the same field of endeavor of generating the re-sampling the data.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to have modified the operating a neural network using an encoder-based model to provide a time series forecast, the method comprising: such that the first scale resolution is less than a scale resolution of the time series dataset; processing as a first iteration, using the model, the initial input to generate a first output; upsampling by an upsampling function the first output to generate a second input having a second scale resolution, taught by KO, to include , the second scale resolution being higher than the first scale resolution , taught by Viswanathan. The modification would have been obvious because one of the ordinary skills in art would be motivated to generate the high resolution image data, (Viswanathan, [par.0052], “In an example embodiment, the scale of a low resolution image may refer to the up-sampling factor required to generate a higher resolution image of the pre-defined and/or configurable resolution and/or size based on the low resolution image. For example, the up-sampling network may be trained to generate higher resolution images of a pre-defined and/or configurable resolution and/or size independent of the starting resolution and/or size of the low resolution image. For example, the up-sampling network may be trained to receive a first low resolution image at a first scale and use a first set of network weights and/or parameters to generate a first higher resolution image of the pre-defined and/or configurable resolution and/or size and to receive a second low resolution image at a second scale and use a second set of network weights and/or parameters to generate a second higher resolution image of the pre-defined and/or configurable resolution and/or size. The up-sampling network may be trained to receive a third low resolution image at a third scale that is between the first scale and the second scale and to generate a third higher resolution image of the pre-defined and/or configurable resolution and/or size using the a third set of network weights and/or parameters that is determined by (e.g., non-linearly) interpolating between the first and second set of network weights and/or parameters. An interpolation function and the corresponding weights and/or parameters may be determined and/or learned based on the loss function and/or error-weight relationship at the different scales..”.)).
Regarding claim 10, Ko teaches the method of claim 1, wherein the model is a probabilistic model (Ko, [Par.0056-0058], “As described later, the model in the first embodiment is characterized in that an attention mechanism for calculating a time attention is incorporated into each of the downsampling blocks 300 and the upsampling blocks 310…[0057] The model processes the time-series data on the wave, to thereby output such time-series data on a probability as shown in FIG. 3B, which indicates whether or not the target wave has arrived for each time step. The horizontal axis indicates the time step, and the vertical axis indicates the probability that the target wave has arrived. When the target wave has arrived, the probability is 1.”).
Regarding claim 11, KO teaches an artificial neural network operated in accordance with the method of claim 1 (Ko, [Par.0060-0061], “The downsampling block 300 includes a one-dimensional convolutional layer 400, two one-dimensional residual blocks 401, and a one-dimensional max pooling layer 402. The structure of the downsampling block 300 illustrated in FIG. 4A is merely an example, and is not limited thereto. Some of the components may be excluded, or another component may be included. In addition, the order of input and output may be changed.[0061] The upsampling block 310 includes a one-dimensional upsampling layer 403, a connected layer 404, a one-dimensional convolutional layer 400, and two one-dimensional residual blocks 401. The structure of the upsampling block 310 illustrated in FIG. 4B is merely an example, and is not limited thereto. Some of the components may be excluded, or another component may be included. In addition, the order of input and output may be changed.”).
Regarding claim 12, Ko teaches a system comprising: a processor;a database storing a time series dataset that is communicatively coupled to the processor; and a memory that is communicatively coupled to the processor and that has stored thereon computer program code that is executable by the processor and that, when executed by the processor, causes the processor to retrieve the time series dataset from the database and to use the time series dataset to perform the method of claim 1 (KO, [par.0115], “It may also be possible that the program codes of the software that realizes the functions of the embodiment are stored on storing means such as a hard disk or a memory of the computer or on a storage medium such as a CD-RW or a CD-R by distributing the program codes through a network and that the CPU that the computer is provided with reads and executes the program codes stored on the storing means or on the storage medium.”.).
Claims 2, 3, 13, 14 are rejected under 35 U.S.C. 103 as being unpatentable over KO et al. (PUB. No 20220187486-hereinafter, KO) in view of Viswanathan et al. (PUB. No 20180293706-hereinafter, Viswanathan) and further in view of YAMADA et al. (PUB. No 20220196615-hereinafter, YAMADA).
Regarding claim 2, KO teaches the method of claim 1 further comprising continuing to iterate using one or more subsequent iterations using the model and the upsampling function (KO, [Par.0052, 0059-0061], “FIG. 4A is a diagram for illustrating an example of a structure of the downsampling block 300 in the first embodiment. FIG. 4B is a diagram for illustrating an example of a structure of the upsampling block 310 in the first embodiment. [0060] The downsampling block 300 includes a one-dimensional convolutional layer 400, two one-dimensional residual blocks 401, and a one-dimensional max pooling layer 402. The structure of the downsampling block 300 illustrated in FIG. 4A is merely an example, and is not limited thereto. Some of the components may be excluded, or another component may be included. In addition, the order of input and output may be changed. [0061] The upsampling block 310 includes a one-dimensional upsampling layer 403, a connected layer 404, a one-dimensional convolutional layer 400, and two one-dimensional residual blocks 401. The structure of the upsampling block 310 illustrated in FIG. 4B is merely an example, and is not limited thereto. Some of the components may be excluded, or another component may be included. In addition, the order of input and output may be changed.” Examiner’s note, the second iteration (decoding operation) is implementing on the first output or the second input is inputting into the second iteration (decoding operation), wherein, decoding operation for outputting data for predicting the first motion time of the target wave.).
However, Ko does not teach continuing to iterate using one or more subsequent iterations using the model and the upsampling function until a resolution scale of the time series forecast matches the scale resolution of the time series dataset,
On the other hand, YAMADA teaches continuing to iterate using one or more subsequent iterations using the model and the upsampling function until a resolution scale of the time series forecast matches the scale resolution of the time series dataset (YAMADA, [Par.0046], “The image generation unit 21 creates a chromatogram based on the chromatogram waveform data as a time-series signal, and converts the chromatogram waveform (chromatogram curve) indicating a change in signal intensity over time into a two-dimensional image having a pixel, the number of which is predetermined (step S2). Here, the number of the pixels is, as an example, 512×512. When being converted into the image, the chromatogram waveform is standardized in size in a Y direction such that a peak top of a peak, which is the greatest in signal intensity among the peaks in the chromatogram waveform, matches an upper side of the image of a rectangular shape. Concurrently, the chromatogram waveform is standardized in size in an X direction such that an entire range of measurement time or a part of the entire range of measurement time (e.g., a range of measurement time specified by the user) matches a length of the image of the rectangular shape in the X direction (a horizontal direction) (step S3). Note that, when the chromatogram waveform is standardized in size in the X direction and when the data point is less than 512 pixels, the chromatogram waveform data may be appropriately up-sampled and converted into a high-resolution waveform in accordance with the original chromatogram waveform data.”).
Ko, Viswanathan and YAMADA are analogous in arts because they have the same field of endeavor of generating the sampling the data.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to have modified the continuing to iterate using one or more subsequent iterations using the model and the upsampling function, taught by KO, to include the continuing to iterate using one or more subsequent iterations using the model and the upsampling function until a resolution scale of the time series forecast matches the scale resolution of the time series dataset, taught by YAMADA. The modification would have been obvious because one of the ordinary skills in art would be motivated to improve the work efficiency of the operation, (YAMADA, [Par.0107], “With the analyzer according to the fifth aspect of the present invention, data to be rechecked is efficiently extracted, which further improves work efficiency of the operator.”).
Regarding claim 3, Ko teaches the method of claim 1, wherein a resolution scale of the time series (KO, [Par. 0055] In FIG. 2, the numbers by the solid arrows each represent numbers of time steps and channels of input or output data. For example, the time-series data on the wave having a time step of 10,000 is input to the downsampling block 300 in the first tier, and eight feature maps having a time step of 2,000 (the number of channels being 8) are output.”);
However, Ko does not teach wherein a resolution scale of the time series forecast matches the scale resolution of the time series dataset,
On the other hand, YAMADA teaches wherein a resolution scale of the time series forecast matches the scale resolution of the time series dataset (YAMADA, [Par.0046], “The image generation unit 21 creates a chromatogram based on the chromatogram waveform data as a time-series signal, and converts the chromatogram waveform (chromatogram curve) indicating a change in signal intensity over time into a two-dimensional image having a pixel, the number of which is predetermined (step S2). Here, the number of the pixels is, as an example, 512×512. When being converted into the image, the chromatogram waveform is standardized in size in a Y direction such that a peak top of a peak, which is the greatest in signal intensity among the peaks in the chromatogram waveform, matches an upper side of the image of a rectangular shape. Concurrently, the chromatogram waveform is standardized in size in an X direction such that an entire range of measurement time or a part of the entire range of measurement time (e.g., a range of measurement time specified by the user) matches a length of the image of the rectangular shape in the X direction (a horizontal direction) (step S3). Note that, when the chromatogram waveform is standardized in size in the X direction and when the data point is less than 512 pixels, the chromatogram waveform data may be appropriately up-sampled and converted into a high-resolution waveform in accordance with the original chromatogram waveform data.”).
Ko, Viswanathan and YAMADA are analogous in arts because they have the same field of endeavor of generating the sampling the data.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to have modified the a resolution scale of the time series, taught by KO, to include the resolution scale of the time series forecast matches the scale resolution of the time series dataset, taught by YAMADA. The modification would have been obvious because one of the ordinary skills in art would be motivated to improve the work efficiency of the operation, (YAMADA, [Par.0107], “With the analyzer according to the fifth aspect of the present invention, data to be rechecked is efficiently extracted, which further improves work efficiency of the operator.”).
Regarding claim 13 is rejected for the same reason as the claim 2, since these claims recite the same limitations.
Regarding claim 14 is rejected for the same reason as the claim 3, since these claims recite the same limitations.
Claims 4, 8, 9, 10, 15, 19, 20 are rejected under 35 U.S.C. 103 as being unpatentable over KO et al. (PUB. No 20220187486-hereinafter, KO) in view of Viswanathan et al. (PUB. No 20180293706-hereinafter, Viswanathan) and further in view of Mentl et al. (PUB. No 20180240219-hereinafter, Mentl).
Regarding claim 4, KO teaches the method of claim 1 further comprising using the encoder but it does not teach using a same encoder for each of the first iteration and the second iteration,
On the other hand, Mentl teaches using a same encoder for each of the first iteration and the second iteration (Mentl, [Par.0038], “[0038], Referring back to FIG. 1, the denoising autoencoder 100 is trained on patches of the noisy image I. as input 101. It follows that the decomposition block 105, thresholding block 107 and reconstruction block 109 are trained for each patch (e.g., repeats training for each patch) for a plurality of training images in an training dataset.”).
Ko, Viswanathan and Menlt are analogous in arts because they have the same field of endeavor of generating the sampling the data.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to have modified the encoder, taught by KO, to include using a same encoder for each of the first iteration and the second iteration, taught by Mentl. The modification would have been obvious because one of the ordinary skills in art would be motivated to improve the process of denoising the images, (Mentl, [Par.0028], “raditional deep learning approaches suffer from slow convergence, making traditional approaches inapplicable for real-time applications (i.e., interventional surgical procedures and the like). Further, by initializing parameters of the Dictionary D for a given type of dataset, the training results only learn a few coefficients, preventing the trained network from fully adapting to the particular class of image datasets provided. The present embodiments provide an approach of deep learning that overcomes the drawbacks of the traditional deep learning approaches by unfolding the thresholding iterations into independently trainable and randomly initialized layers, such as using a network multiscale denoising autoencoders or another deep learning networks. Accordingly, a multiscale patch-based sparse image representation is learned and applied for denoising image data to reduce the computational expense for denoising the image data. Reducing the computational expense improves image processor speeds, allowing for real-time scanning and denoising. Further, more accurate images may be reconstructed and displayed, providing for better treatment and diagnosis with lower radiation doses used to increase patient safety.”).
Regarding claim 8, Ko as modified in view of Menlt teaches the method of claim 1 further comprising using a loss function on the second output in order to quantify a error present in the time series forecast, (Ment, [Par.0008], “The method further includes upsampling the denoised image data obtained at the second scale back to the first scale and applies a learnt linear filter to denoised image data obtained at the first scale and the denoised image data upsampled from the second scale. A summation is performed on the filtered denoised data to obtain final denoised image data. The weights of the deep neural networks are randomly initialized during training, and the method compares the final denoised image with target image data to update the weights of the first and second deep neural networks using backpropagation. The trained deep neural networks are stored as a as a deep-learning based network.” And [Par.0034], “Referring to FIG. 1, the denoising autoencoder 100 updates network parameters based on gradient backpropagation with respect to the mean squared error (MSE) or another loss between the reconstructed output 103 and the input 101 of the network 100.” ).
Ko, Viswanathan and Mentl are analogous in arts because they have the same field of endeavor of generating the sampling the data.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to have modified the method of the claim 1, taught by KO, to include using a loss function on the second output in order to quantify a error present in the time series forecast, taught by Mentl. The modification would have been obvious because one of the ordinary skills in art would be motivated to improve the process of denoising the images, (Mentl, [Par.0028], “traditional deep learning approaches suffer from slow convergence, making traditional approaches inapplicable for real-time applications (i.e., interventional surgical procedures and the like). Further, by initializing parameters of the Dictionary D for a given type of dataset, the training results only learn a few coefficients, preventing the trained network from fully adapting to the particular class of image datasets provided. The present embodiments provide an approach of deep learning that overcomes the drawbacks of the traditional deep learning approaches by unfolding the thresholding iterations into independently trainable and randomly initialized layers, such as using a network multiscale denoising autoencoders or another deep learning networks. Accordingly, a multiscale patch-based sparse image representation is learned and applied for denoising image data to reduce the computational expense for denoising the image data. Reducing the computational expense improves image processor speeds, allowing for real-time scanning and denoising. Further, more accurate images may be reconstructed and displayed, providing for better treatment and diagnosis with lower radiation doses used to increase patient safety.”).
Regarding claim 9, KO teaches the model but it does not teach the model is a transformer model,
On the other hand, Mentl teaches wherein the model is a transformer model (Ko, [Par.0079], “The implementation example of FIG. 8B is described. The time attention block 503 inputs the feature map to a plurality of convolutional layers having a pyramid structure and different scales to calculate attentions (feature maps). In this case, three convolutional layers of 1×1, 1×5, and 1×5 are used. The time attention block 503 connects the attentions for the scales to one another, and inputs the connected attentions to a one-dimensional convolutional layer. The time attention block 503 multiplies the feature map before the conversion by the attention (the feature map) output from the one-dimensional convolutional layer to output a feature map to which the time attention is added.” ).
Ko, Viswanathan and Mentl are analogous in arts because they have the same field of endeavor of generating the sampling the data.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to have modified the model, taught by KO, to include the model is a transformer model, taught by Mentl. The modification would have been obvious because one of the ordinary skills in art would be motivated to improve the process of denoising the images, (Mentl, [Par.0028], “traditional deep learning approaches suffer from slow convergence, making traditional approaches inapplicable for real-time applications (i.e., interventional surgical procedures and the like). Further, by initializing parameters of the Dictionary D for a given type of dataset, the training results only learn a few coefficients, preventing the trained network from fully adapting to the particular class of image datasets provided. The present embodiments provide an approach of deep learning that overcomes the drawbacks of the traditional deep learning approaches by unfolding the thresholding iterations into independently trainable and randomly initialized layers, such as using a network multiscale denoising autoencoders or another deep learning networks. Accordingly, a multiscale patch-based sparse image representation is learned and applied for denoising image data to reduce the computational expense for denoising the image data. Reducing the computational expense improves image processor speeds, allowing for real-time scanning and denoising. Further, more accurate images may be reconstructed and displayed, providing for better treatment and diagnosis with lower radiation doses used to increase patient safety.”).
Regarding claim 15 is rejected for the same reason as the claim 4, since these claims recite the same limitations.
Regarding claim 19 is rejected for the same reason as the claim 8, since these claims recite the same limitations.
Regarding claim 20 is rejected for the same reason as the claim 9, since these claims recite the same limitations.
Claims 5, 16 are rejected under 35 U.S.C. 103 as being unpatentable over KO et al. (PUB. No 20220187486-hereinafter, KO) in view of Viswanathan et al. (PUB. No 20180293706-hereinafter, Viswanathan) and further in view of Sun et al. (PUB. No 20230386500-hereinafter, Sun).
Regarding claim 5, KO teaches the method of claim 1 further comprising encoder but it does not teach using a different encoder for each of the first iteration and the second iteration,
However, it does not teach using a different encoder for each of the first iteration and the second iteration (SUn, [Par.006], “[0006] he multi-scale nested block may comprise a first encoding layer configured to generate a first encoded data set by performing a convolution based on the input data. The multi-scale nested block may comprise a second encoding layer configured to generate a second encoded data set by performing a convolution based on the first downsampled input data set.” [0008] In the multi-scale nested block, the encoding and convolutional layers may be identical or different.).
Ko, Viswanathan and Sun are analogous in arts because they have the same field of endeavor of generating the sampling the data.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to have modified the encoder, taught by KO, to include the resolution scale of the using a different encoder for each of the first iteration and the second iteration, taught by Sun. The modification would have been obvious because one of the ordinary skills in art would be motivated to optimize the problem, (Sun, [Par.0010], “The third convolutional layer may be also denoted/regarded as first decoding layer, and the third output data set may be denoted as first decoded data set. In other words, the first decoded data set may represent a decoded data set which is at the same scale as the input data. Analogously, the second convolutional layer may be also denoted as second decoding layer, and the second output data set may be denoted as second decoded data set. That is, the second decoded data set may represent a decoded data set which is at a lower scale compared to the input data, or more precisely: at the scale of first downsampled input data set. Thus, according to the forgoing interpretation of the described CNN architecture, the first convolutional layer is coupled in between the two encoding layers and the two decoding layers and may, as a consequence, be also denoted as nested (or intermediate) convolutional layer. As such, the presence of this nested convolutional layer enables increased communication within the proposed CNN architecture. In particular, the introduction of the nested convolutional layer brings the semantic level of the encoded data sets (e.g. encoder feature maps) closer to the semantic level of decoded data sets (e.g. decoder feature maps). The technical advantage is that an optimizer may face an easier optimization problem when the received encoded data sets and the corresponding decoded data sets are semantically more similar.”).
Regarding claim 16 is rejected for the same reason as the claim 5, since these claims recite the same limitations.
Claims 6, 7, 17, 18 are rejected under 35 U.S.C. 103 as being unpatentable over KO et al. (PUB. No 20220187486-hereinafter, KO) in view of Viswanathan et al. (PUB. No 20180293706 -hereinafter, Viswanathan) and further in view of Daniell et al. (PUB. No 20220284269-hereinafter, Daniell).
Regarding claim 6, KO teaches the method of claim 1, but it does not teach using a normalization function on the initial input in order to normalize the initial input before said processing using the model,
On other hand, Daniell teaches using a normalization function on the initial input in order to normalize the initial input before said processing using the model, (Daniell, [par.0118], ‘In some examples, such historical time-series input data may be preprocessed before being input to the first and second models. For example, such historical time-series input data may be down-sampled to an hourly time resolution, padded, normalized (e.g., z-score normalized), etc. Such historical time-series data may be measured over a window of time (e.g., a window of 14 days, etc.) (0089-0100], “Input 301 may undergo preprocessing before being input to model 300. For example, a computing device (e.g., computing device 140) may downsample such input data from a higher time resolution to a lower time resolution representation, for instance hourly (e.g., from multiple data points per hour to one data point per input value per hour). Additionally or alternatively, input 301 may be padded, and/or normalized before entering model 300. For example, input 301 may be z-score normalized before being forwarded into model 300.[0090] Input 301 may thus represent multiple variables at each of a plurality of time steps. For example, input 301 may indicate values for a given set of input variables for each hour of a twelve hour window…[0092], “As an example, input data for 12 time steps, corresponding to a 12-hour time window, may be input to model 300. For each time step (e.g., an hour), there may be an input value for some, or all of the variables described above. This input data (e.g., input 301) flows first through input layer 302, and then through the hidden layers, LSTM layers 302, 304, 306, 308, etc., and finally through output layer 310. Each layer may transform values received from the previous layer. Output layer 310 may calculate an output value for each hour-long timestep. Thus, in accordance with this example, there may be 12 output values, one for each hour. Each such output value may comprise an output vector that indicates fermentation state probabilities during that time window for each known fermentation state.” Examiner’ snote, the plurality of first time series data sets include the time series data is downsampling each hour window before it input into the model.).
Ko and Daniell are analogous in arts because they have the same field of endeavor of generating the time series data.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to have modified the method of claim 1, taught by KO, to include the using a normalization function on the initial input in order to normalize the initial input before said processing using the model, taught by Daniell. The modification would have been obvious because one of the ordinary skills in art would be motivated to the model accuracy, (Daniell, [Par.0098], “During such a training phase, ground truth data associated with known fermentation states may be input to models 410, 440. For example, model 400 may be trained by inputting data from two known similar fermentation states, or two known dissimilar fermentation states into model 400. Model 400 may output a numeric representation of a similarity between the two inputs. Based on whether the numeric similarity output by model 400 for two inputs is correct, the weights and/or similarity function of model 400 may be adjusted to produce more accurate predictions based on minimizing a cost function and/or backpropagation. Such a cost function may be based on mean-squared error in some examples. In some examples, such a cost function may be minimized by algorithms such as gradient descent, stochastic gradient descent, or the Adam optimization algorithm, etc. A cost function may be minimized (e.g., optimized) in various other manners as well. According to some examples, model 400 may be configured to perform gradient clipping when optimizing such a cost function. Gradient clipping may force gradient values not to exceed a specific maximum or minimum value, which may improve model accuracy.”).
Regarding claim 7, Ko teaches the method of the claim 1, but it does not teach using a normalization function on the second input in order to normalize the second input before said processing using the model,
On the other hand, Daniell teaches using a normalization function on the second input in order to normalize the second input before said processing using the model (Daniell, [par.0118], ‘In some examples, such historical time-series input data may be preprocessed before being input to the first and second models. For example, such historical time-series input data may be down-sampled to an hourly time resolution, padded, normalized (e.g., z-score normalized), etc. Such historical time-series data may be measured over a window of time (e.g., a window of 14 days, etc.) (0089-0100], “Input 301 may undergo preprocessing before being input to model 300. For example, a computing device (e.g., computing device 140) may downsample such input data from a higher time resolution to a lower time resolution representation, for instance hourly (e.g., from multiple data points per hour to one data point per input value per hour). Additionally or alternatively, input 301 may be padded, and/or normalized before entering model 300. For example, input 301 may be z-score normalized before being forwarded into model 300.[0090] Input 301 may thus represent multiple variables at each of a plurality of time steps. For example, input 301 may indicate values for a given set of input variables for each hour of a twelve hour window…[0092], “As an example, input data for 12 time steps, corresponding to a 12-hour time window, may be input to model 300. For each time step (e.g., an hour), there may be an input value for some, or all of the variables described above. This input data (e.g., input 301) flows first through input layer 302, and then through the hidden layers, LSTM layers 302, 304, 306, 308, etc., and finally through output layer 310. Each layer may transform values received from the previous layer. Output layer 310 may calculate an output value for each hour-long timestep. Thus, in accordance with this example, there may be 12 output values, one for each hour. Each such output value may comprise an output vector that indicates fermentation state probabilities during that time window for each known fermentation state.” Examiner’s note, the plurality of first time series data sets include the time series data is downsampling each hour window before it input into the model.).
Ko and Daniell are analogous in arts because they have the same field of endeavor of generating the time series data.
Accordingly, it would have been obvious to one of the ordinary skills in the art before the effective filing date of the claimed invention to have modified the method of claim 1, taught by KO, to include the using a normalization function on the second input in order to normalize the second input before said processing using the model, taught by Daniell. The modification would have been obvious because one of the ordinary skills in art would be motivated to the model accuracy, (Daniell, [Par.0098], “During such a training phase, ground truth data associated with known fermentation states may be input to models 410, 440. For example, model 400 may be trained by inputting data from two known similar fermentation states, or two known dissimilar fermentation states into model 400. Model 400 may output a numeric representation of a similarity between the two inputs. Based on whether the numeric similarity output by model 400 for two inputs is correct, the weights and/or similarity function of model 400 may be adjusted to produce more accurate predictions based on minimizing a cost function and/or backpropagation. Such a cost function may be based on mean-squared error in some examples. In some examples, such a cost function may be minimized by algorithms such as gradient descent, stochastic gradient descent, or the Adam optimization algorithm, etc. A cost function may be minimized (e.g., optimized) in various other manners as well. According to some examples, model 400 may be configured to perform gradient clipping when optimizing such a cost function. Gradient clipping may force gradient values not to exceed a specific maximum or minimum value, which may improve model accuracy.”).
Regarding claim 17 is rejected for the same reason as the claim 6, since these claims recite the same limitations.
Regarding claim 18 is rejected for the same reason as the claim 7, since these claims recite the same limitations.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to EM N TRIEU whose telephone number is (571)272-5747. The examiner can normally be reached on Mon-Fri from 9:00-5:00.
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, Omar Fernandez Rivas can be reached on (571) 272-2589. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/E.T./Examiner, Art Unit 2128
/OMAR F FERNANDEZ RIVAS/Supervisory Patent Examiner, Art Unit 2128
Prosecution Insights