DETAILED ACTION
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 .
Response to Arguments
Applicant’s arguments, see Remarks, filed 07/13/2026 have been fully considered:
Regarding applicant's arguments filed with respect to the prior art rejections have been fully considered but they are moot. Applicant has amended the claims to recite new combinations of limitations. Please see below for new grounds of rejection, necessitated by Amendment.
Examiner notes:
Regarding 101 abstract idea, claim 1 is tied to optical communications and the “model update of the trained model step by step based on the data”, improves signal processing performance, as cited in the instant specification at least ¶51, “the model update unit 122 updates the analysis model to gradually adapt the analysis model to each influence factor which affects the data distribution of the transponder. Therefore, even in a case where there is a large divergence between the data acquired in the environment where the base model is generated and the data acquired in the environment of the customer X, the model update is performed so as to reduce the discrepancy step by step, thus enabling the optimization of the analysis model used at a site of installation.”
Claim Rejections - 35 USC § 103
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.
Claim(s) 1-12 is/are rejected under 35 U.S.C. 103 as being unpatentable over XU et al. (US 20220239371 A1) in view of Long et al. (“Deep Transfer Learning with Joint Adaptation Networks”, PMLR 70, 2017) in further in view of Kolouri et al. (US 20210192363 A1).
Regarding claim 1.
Xu teaches a model optimization device for a parameter estimation concerning optical communications, the model optimization device comprising: a memory storing instructions; and one or more processors configured to execute the instructions (see ¶ 35, “the term “circuitry” may refer to one or more or all of the following: (a) hardware-only circuit implementations (such as implementations in only analog and/or digital circuitry) and (b) combinations of hardware circuits and software, such as (as applicable): (i) a combination of analog and/or digital hardware circuit(s) with software/firmware and (ii) any portions of hardware processor(s) with software (including digital signal processor(s)), software, and memory(ies) that work together to cause an apparatus, such as a mobile phone or server, to perform various functions) and (c) hardware circuit(y) and or processor(s)”) to:
acquire see ¶ 40, “In the training stage, a given learning model may be trained by using a large scale of training data, and continuously iterating and updating values of a parameter set of the model until the model can reach a desired objective. After the training, the values of the parameter set of the machine learning model are determined. In the test stage, test samples may be used to test the trained learning model to determine the performance of the learning model. In the application stage, real-world input data may be applied to the trained learning model. The learning model may process the real-world input data based on the parameter set obtained by the training to provide a corresponding output.”, also see ¶ 86-89, i.e. trained model acquired);
acquire data from a terminal device (see ¶ 113, “the feedback information fed back by one or more ONUs 120 to the OLT 110 after performing the training of the neural network 130 may also include part or all of the training data used for training the neural network 130. The training data here includes the pilot sequences received by the ONUs 120 from the OLT 110 via their respective communication channels. In this way, the OLT 110 may obtain real training data from the ONU 120 to perform the update of the initial parameter value set of the neural network 130 more effectively.”, i.e. receiving data from ONU);
generate an updated model by performing a model update of the trained model see ¶ 90, “At the OLT 110, the OLT 110 receives 520 the feedback information from the ONU 120, and updates 525, based on the feedback information, the maintained initial parameter value set for the neural network 130. For example, the parameter update module 320 in the OLT 110 is configured to update the initial parameter value set of the neural network 130. The initial parameter value set currently maintained by the OLT 110 is sometimes herein referred to as “a first initial parameter value set”, and the updated initial parameter value set is sometimes herein referred to as “a second initial parameter value set”.”, i.e. updated model from data, also see ¶ 119, “The total gradient θ of the whole neural network may be updated iteratively in a plurality of training steps. After the iterative update reaches the expected objective, for example, the iteration number reaches a threshold, or the loss value of the neural network is less than a predetermined threshold, the update of the initial parameter value set of the neural network is completed.”, i.e. step by step update );
and output the updated model to an output destination device corresponding to the terminal device (see ¶ 96, “after completing the update of the initial parameter value set, the OLT 110 transmits the updated initial parameter value set (i.e., the “second initial parameter value set”) to one or more further ONUs 120. The further ONUs 120 may, for example, be new ONUs 120 that are to establish communication channels with the OLT 110. Those ONUs 120 may use the updated initial parameter value set to train the neural networks 130 deployed, in order to use the trained neural networks 130 to process signals received from the OLT 110.”).
Xu do not specifically teach acquire a trained model and update of the trained model step by step, based on the data, such that the trained model is adapted one by one in order to a plurality of different influence factors that affect a data distribution of the terminal device.
Long teaches acquire a trained model and update of the trained model step by step based on the data (see page 1, introduction, “Transfer learning becomes more challenging when domains may change by the joint distributions of input features and output labels, which is a common scenario in practical ap plications. First, deep networks generally learn the complex function from input features to output labels via multilayer feature transformation and abstraction. Second, deep features in standard CNNs eventually transition from general to specific along the network, and the transferability of features and classifiers decreases when the cross-domain discrepancy increases”, also see page 3, section 4, “we opt to fine-tune the features of convolutional layers when transferring pretrained deep models from source domain to target domain.”).
Both Xu and Long pertain to the problem of model learning optimization, thus being analogous. It would have been obvious to one skilled in the art before the effective filing date of the claimed invention to combine Xu and Long to teach the above limitations. The motivation for doing so would be “we present joint adaptation networks (JAN), which learn a transfer network by aligning the joint distributions of multiple domain-specific layers across domains based on a joint maximum mean discrepancy (JMMD) criterion. Adversarial training strategy is adopted to maximize JMMD such that the distributions of the source and target domains are made more distinguishable. Learning can be performed by stochastic gradient descent with the gradients computed by back-propagation in linear-time. Experiments testify that our model yields state of the art results on standard datasets..” (see Long abstract).
Xu and Long do not specifically teach acquire a trained model and update of the trained model step by step, based on the data, such that the trained model is adapted one by one in order to a plurality of different influence factors that affect a data distribution of the terminal device.
Kolouri teaches acquire a trained model and update of the trained model step by step, based on the data, such that the trained model is adapted one by one in order to a plurality of different influence factors that affect a data distribution of the terminal device (see ¶ 7, “The new task data and a plurality of past learned tasks is forced to share a data distribution in an embedding space, resulting in a shared generative data distribution. Using the shared generative data distribution, a set of pseudo-data points is generated for the past learned tasks. Each new domain is learned using both the set of pseudo-data points and the new task data. The machine learning model is updated using both the set of pseudo-data points and the new task data.”, also see ¶ 66-69, “the trained system is able to generate pseudo-data points 303 for the past tasks 310 (i.e., classifying input sensory data). To this end, the current tasks 305 (denoted as t+1 and t+2) and past tasks 310 are coupled by enforcing them to share the same parametric distribution in a task-invariant embedding space 304. This shared distribution can then be used to generate pseudo-data points 303 using the decoder network 314 that can be used for experience replay (in the second step). In a third step, the new domain is learned through matching its distribution in the embedding space 304. In a fourth step, the new learned knowledge is used to update the embedding distribution. The method enables a machine/agent to remember previously learned tasks (i.e., past tasks 310) and easily learn new tasks (i.e., current tasks 305) without corrupting the knowledge of previously learned tasks (i.e., past tasks 310). An “agent” is any type of machine or robot (autonomous platform) that learns the tasks.”, also see ¶ 71-73, “The goal is to use the encoded distribution in the embedding space 304 to expand the concepts that are captured in the embedding space 304 such that catastrophic forgetting does not occur. The gist of the idea is to update the encoder network 306 such that each subsequent task is learned so that its distribution in the embedding space 304 matches the distribution that is shared by {Z.sup.(t)}.sub.t=1.sup.T−1 at t=T. Since this distribution is initially learned via Z.sup.(t) and subsequent tasks (with the same underlying object categories or concepts) are enforced to share this distribution in the embedding space 304 with Z.sup.(t), it does not need to learn it from scratch, as the concepts are shared across the tasks (e.g., symbols of digits in different language scripts that allude to the same concepts of numerosity).”, also see ¶ 75-77, and 114-117 teaches autonomous vehicles adapting to new environments and personalized requirements).
Xu, Long and Kolouri pertain to the problem of model learning optimization, thus being analogous. It would have been obvious to one skilled in the art before the effective filing date of the claimed invention to combine Xu, Long and Kolouri to teach the above limitations. The motivation for doing so would be “continual adaptation of a machine learning model implemented in an autonomous platform. The system adapts knowledge previously learned by the machine learning model for performance in a new domain. The system receives a consecutive sequence of new domains comprising new task data. The new task data and past learned tasks are forced to share a data distribution in an embedding space, resulting in a shared generative data distribution. The shared generative data distribution is used to generate a set of pseudo-data points for the past learned tasks. Each new domain is learned using both the set of pseudo-data points and the new task data. The machine learning model is updated using both the set of pseudo-data points and the new task data… autonomous self-driving cars need to perform well in different weather conditions and types of roads, despite being trained on more limited conditions. Model retraining is not a feasible solution for lifelong learning because collecting labeled data to supervise learning is time-consuming and computationally expensive. As a result, a lifelong learning system should be able to explore and learn the new condition fast using a minimal number of labeled data points without forgetting what has been learned before to avoid computationally expensive model retraining” (see Kolouri abstract and ¶ 4).
Regarding claim 2.
Xu, Long and Kolouri teaches the model optimization device according to claim 1,
Xu further teaches wherein the processor generates the updated model by performing, step by step, the model update which adapts the trained model to a different terminal device and the model update which adapts the trained model to a different environment (see ¶ 115, “The OLT 110 may collect real training data used by the plurality of ONUs 120 respectively, to update the initial parameter value set of the neural network 130. In some embodiments, instead of using only training data from one ONU 120 or mixing up training data from different ONUs 120, the OLT 110 may train the neural network 130 through a Model Agnostic Meta Learning (MAML) process. ”, i.e. different terminal devices, also see ¶ 116, “At the OLT 110, common features learned by the neural network 130 at different ONUs 120 may be obtained through the MAML, so that the updated initial parameter value set can better facilitate the neural network 130 to quickly adapt to a new environment, i.e., to train the neural network 130 at a new ONU 120 more quickly to meet an expected objective.”, i.e. different environments also see ¶ 119, “The total gradient θ of the whole neural network may be updated iteratively in a plurality of training steps. After the iterative update reaches the expected objective, for example, the iteration number reaches a threshold, or the loss value of the neural network is less than a predetermined threshold, the update of the initial parameter value set of the neural network is completed”, i.e. step by step update).
Long further teaches the step-by-step adaption (see page 1, introduction, “Transfer learning becomes more challenging when domains may change by the joint distributions of input features and output labels, which is a common scenario in practical ap plications. First, deep networks generally learn the complex function from input features to output labels via multilayer feature transformation and abstraction. Second, deep features in standard CNNs eventually transition from general to specific along the network, and the transferability of features and classifiers decreases when the cross-domain discrepancy increases”, also see page 3, section 4, “we opt to fine-tune the features of convolutional layers when transferring pretrained deep models from source domain to target domain.”)
The motivation utilized in the combination of claim 1, super, applies equally as well to claim 2.
Regarding claim 3.
Xu, Long and Kolouri teaches the model optimization device according to claim 2,
Long further teaches wherein the processor performs the model update which adapts to the different environment after the model update which adapts to the different terminal device (see page 1, introduction, “Transfer learning becomes more challenging when domains may change by the joint distributions of input features and output labels, which is a common scenario in practical ap plications. First, deep networks generally learn the complex function from input features to output labels via multilayer feature transformation and abstraction. Second, deep features in standard CNNs eventually transition from general to specific along the network, and the transferability of features and classifiers decreases when the cross-domain discrepancy increases”, also see page 3, section 4, “we opt to fine-tune the features of convolutional layers when transferring pretrained deep models from source domain to target domain.”)
The motivation utilized in the combination of claim 1, super, applies equally as well to claim 3.
Regarding claim 4.
Xu, Long and Kolouri teaches the model optimization device according to claim 1,
Xu further teaches wherein the processor generates the updated model by performing, step by step, the model update which adapts to a different terminal device, the model update which adapts the trained model to a different network, and the model update which adapts the trained model to different states (see ¶ 115, “The OLT 110 may collect real training data used by the plurality of ONUs 120 respectively, to update the initial parameter value set of the neural network 130. In some embodiments, instead of using only training data from one ONU 120 or mixing up training data from different ONUs 120, the OLT 110 may train the neural network 130 through a Model Agnostic Meta Learning (MAML) process. ”, i.e. different terminal devices, also see ¶ 116, “At the OLT 110, common features learned by the neural network 130 at different ONUs 120 may be obtained through the MAML, so that the updated initial parameter value set can better facilitate the neural network 130 to quickly adapt to a new environment, i.e., to train the neural network 130 at a new ONU 120 more quickly to meet an expected objective.”, i.e. different environments also see ¶ 119, “The total gradient θ of the whole neural network may be updated iteratively in a plurality of training steps. After the iterative update reaches the expected objective, for example, the iteration number reaches a threshold, or the loss value of the neural network is less than a predetermined threshold, the update of the initial parameter value set of the neural network is completed”, i.e. step by step update, also see ¶ 49, “there may be different optical fiber lengths, different bandwidths, and the like”, also see ¶ 98-99, “after completing the training of the neural network 130, an ONU 120 may perform a channel activation stage with the OLT 110 to activate the communication channel for a specific wavelength…after activating the communication channel, the ONU 120 may transmit feedback information on the training of the neural network 130 to the OLT 110 over the activated communication channel”).
Long further teaches step by step sequence (see figure 1, page 4, “Since deep features eventually transition from general to specific along the network, activations in multiple domain-specific layers L are not safely transferable. And the joint distributions of the activations P(Zs1,. .., Zs|L|) and Q(Zt1,...,Zt|L|) in these layers should be adapted by JMMD minimization”, also see page 1, introduction, “Transfer learning becomes more challenging when domains may change by the joint distributions of input features and output labels, which is a common scenario in practical ap plications. First, deep networks generally learn the complex function from input features to output labels via multilayer feature transformation and abstraction. Second, deep features in standard CNNs eventually transition from general to specific along the network, and the transferability of features and classifiers decreases when the cross-domain discrepancy increases”, also see page 3, section 4, “we opt to fine-tune the features of convolutional layers when transferring pretrained deep models from source domain to target domain.”)
The motivation utilized in the combination of claim 1, super, applies equally as well to claim 4.
Regarding claim 5.
Xu, Long and Kolouri teaches the model optimization device according to claim 1,
Xu further teaches wherein the processor updates a model so as to adapt an analysis model to unique characteristics of the terminal device, optical network characteristics, an installation state of the terminal device step by step (see ¶ 119, “The total gradient θ of the whole neural network may be updated iteratively in a plurality of training steps. After the iterative update reaches the expected objective, for example, the iteration number reaches a threshold, or the loss value of the neural network is less than a predetermined threshold, the update of the initial parameter value set of the neural network is completed”, i.e. step by step update, also see ¶ 49, “there may be different optical fiber lengths, different bandwidths, and the like”, also see ¶ 109, “the assistance communication apparatus 730 may include a separate communication apparatus, which may be detachably connected to the ONU 120 for communication therewith… for example, be manually installed to the ONU 120, to obtain the initial parameter value set for the ONU 120.”).
Regarding claim 6.
Xu, Long and Kolouri teaches the model optimization device according to claim 4,
Long further teaches wherein the processor first performs the model update which adapts to the different terminal device, next performs the model update which adapts to the different network, and further performs the model update which adapts to the different states (see figure 1, page 4, “Since deep features eventually transition from general to specific along the network, activations in multiple domain-specific layers L are not safely transferable. And the joint distributions of the activations P(Zs1,. .., Zs|L|) and Q(Zt1,...,Zt|L|) in these layers should be adapted by JMMD minimization”, also see page 1, introduction, “Transfer learning becomes more challenging when domains may change by the joint distributions of input features and output labels, which is a common scenario in practical ap plications. First, deep networks generally learn the complex function from input features to output labels via multilayer feature transformation and abstraction. Second, deep features in standard CNNs eventually transition from general to specific along the network, and the transferability of features and classifiers decreases when the cross-domain discrepancy increases”, also see page 3, section 4, “we opt to fine-tune the features of convolutional layers when transferring pretrained deep models from source domain to target domain.”)
The motivation utilized in the combination of claim 1, super, applies equally as well to claim 6.
Regarding claim 7.
Xu, Long and Kolouri teaches the model optimization device according to claim 4,
Xu further teaches wherein the processor generates a generic model based on a plurality of updated models acquired by the model update which adapts to the different states (see ¶ 115, “instead of using only training data from one ONU 120 or mixing up training data from different ONUs 120, the OLT 110 may train the neural network 130 through a Model Agnostic Meta Learning (MAML) process. The MAML may combine learning losses of a plurality of models learned in different environments (also referred to as tasks), and compute the loss with new data to update the model, thereby improving the model robustness.”, also see ¶ 116, “At the OLT 110, common features learned by the neural network 130 at different ONUs 120 may be obtained through the MAML, so that the updated initial parameter value set can better facilitate the neural network 130 to quickly adapt to a new environment, i.e., to train the neural network 130 at a new ONU 120 more quickly to meet an expected objective. In addition, the application of MAML enables the OLT 110 to collect a small amount of real training data from a single ONU 120 to complete the update of the parameter values.”).
Regarding claim 8.
Xu, Long and Kolouri teaches the model optimization device according to claim 4,
Xu further teaches wherein the processor generates a generic model based on a plurality of updated models and a plurality of pieces of data which are acquired by the model update which adapts to the different states (see ¶ 115, “instead of using only training data from one ONU 120 or mixing up training data from different ONUs 120, the OLT 110 may train the neural network 130 through a Model Agnostic Meta Learning (MAML) process. The MAML may combine learning losses of a plurality of models learned in different environments (also referred to as tasks), and compute the loss with new data to update the model, thereby improving the model robustness.”, also see ¶ 117, “training data D and D′ of a plurality of tasks may be randomly selected from the training data corresponding to different tasks.”, also see ¶¶ 115-118, combining losses to update model).
Regarding claim 9.
Xu, Long and Kolouri teaches the model optimization device according to claim 4,
Xu further teaches wherein the processor generates a generic model based on a plurality of updated models acquired by the model update which adapts to the different network (see ¶ 116, “At the OLT 110, common features learned by the neural network 130 at different ONUs 120 may be obtained through the MAML, so that the updated initial parameter value set can better facilitate the neural network 130 to quickly adapt to a new environment, i.e., to train the neural network 130 at a new ONU 120 more quickly to meet an expected objective. In addition, the application of MAML enables the OLT 110 to collect a small amount of real training data from a single ONU 120 to complete the update of the parameter values.”, see ¶ 115, “instead of using only training data from one ONU 120 or mixing up training data from different ONUs 120, the OLT 110 may train the neural network 130 through a Model Agnostic Meta Learning (MAML) process. The MAML may combine learning losses of a plurality of models learned in different environments (also referred to as tasks), and compute the loss with new data to update the model, thereby improving the model robustness.”).
Regarding claim 10.
Xu, Long and Kolouri teaches the model optimization device according to claim 4,
Xu further teaches wherein the processor generates a generic model based on a plurality of updated models acquired by the model update which adapts to the different terminal device (see ¶ 115, “instead of using only training data from one ONU 120 or mixing up training data from different ONUs 120, the OLT 110 may train the neural network 130 through a Model Agnostic Meta Learning (MAML) process. The MAML may combine learning losses of a plurality of models learned in different environments (also referred to as tasks), and compute the loss with new data to update the model, thereby improving the model robustness.”, also see ¶ 116, “At the OLT 110, common features learned by the neural network 130 at different ONUs 120 may be obtained through the MAML, so that the updated initial parameter value set can better facilitate the neural network 130 to quickly adapt to a new environment, i.e., to train the neural network 130 at a new ONU 120 more quickly to meet an expected objective. In addition, the application of MAML enables the OLT 110 to collect a small amount of real training data from a single ONU 120 to complete the update of the parameter values.”).
Claim 11 recites a method to perform the device recited in claim 1. Therefore the rejection of claim 1 above applies equally here.
Claim 12 recites a non-transitory computer readable recording medium storing a program to perform the device recited in claim 1. Therefore the rejection of claim 1 above applies equally here.
Claim(s) claims 13-15 is/are rejected under 35 U.S.C. 103 as being unpatentable over XU et al. (US 20220239371 A1) in view of Long et al. (“Deep Transfer Learning with Joint Adaptation Networks”, PMLR 70, 2017) in further in view of Kolouri et al. (US 20210192363 A1) in view of Helenius et al. (US 20200311585 A1).
Regarding claim 13.
Xu, Long and Kolouri teaches the model optimization device according to claim 1,
Xu, Long and Kolouri do not teach claim 13.
Helenius teaches wherein the processor is further configured to: manage optimization phases for the plurality of different influence factors in a tree structure, and perform the model update in each layer according to a shift command of an optimization phase of the optimization phases (see ¶ 74, “The AP model 613 samples additional sets of N outcome vectors from the product/account feature, outcome data 611 and trains an additional random tree until a desired number of random trees is reached. If the random forest has been previously trained when the AP model 613 receives product/account feature, outcome data 611, the incoming data can be used to update the nodes of each random tree in the random forest as well as the likelihood values at the leaves of each random tree.”, also see ¶ 76, “The RL model 619 is initialized with random or deterministic Q function values. The Q function values can all be initialized to zero or can be generated from a probability distribution. In some embodiments, the Q function values are initialized to high values to encourage the RL model 619 to explore many potential actions from a given state. The first observed reward can be used to reset the initial Q function values. The choice of Q function initialization can vary significantly depending on the structure of the training data. The RL model 619 updates the Q function value at each state/action pair represented by consecutive products in the product sequence/outcome data 617 according to the outcome, using an iteration update that includes an estimate of future rewards. The discount factor of the future reward can be small, such that future rewards are not overvalued compared to current rewards. If the RL model 619 is already initialized, the product sequence/outcome data 617 is used to update the Q function values for each state/action pair.”).
Xu, Long, Kolouri and Helenius pertain to the problem of model learning optimization, thus being analogous. It would have been obvious to one skilled in the art before the effective filing date of the claimed invention to combine Xu, Long, Kolouri and Helenius to teach the above limitations. The motivation for doing so would be “trained AP model assigns propensity values to each product corresponding to received account characteristics. The trained RL model generates an optimal sequence of products that maximizes the reward over future realized opportunities. The target engagement sequence generators create target engagement sequences corresponding to the optimal sequence of products. The recommender prunes the optimal sequence of products based on the propensity values from the trained AP model, the completeness of these target engagement sequences, and a desired product sequence length. The recommender uses the remaining products, validated on three models, for account/product recommendations” (see Helenius abstract).
Claim 14 recites a method to perform the device recited in claim 13. Therefore the rejection of claim 13 above applies equally here.
Claim 15 recites a non-transitory computer readable recording medium storing a program to perform the device recited in claim 13. Therefore the rejection of claim 13 above applies equally here.
Related prior arts:
Flanagan et al. (US 20220083911 A1) teaches a master machine learning model for generating a user recommendation related to use of an application of the user equipment, calculate a model update for the master machine learning model using the master machine learning model and data related to one or more of a user of the user equipment or a user interaction with the user equipment, encode the calculated model update using an ε-differential privacy mechanism and transmit the ε-differential privacy encoded model update.
Duesterwald et al. (US 10452994 B2) teaches One or more processors obtain and deploy a cognitive engine that utilizes artificial intelligence (AI), machine learning, and/or similar algorithms. One or more processors obtain and deploy a version of a trained model that includes data that supports cognitive operations of the cognitive engine within a cognitive service. In response to changes to the input used to produce the trained model, one or more processors obtain and deploy a subsequent version of the trained model in support of the cognitive service.
Zhang et al. (US 20190089463 A1) teaches performing registration of an optical network unit using optical communication that uses an on-off key based modulation, performing, upon completion of the registration, link estimation, and using a multi-level modulation scheme, whose parameters are based on the link estimation, to perform subsequent communication with the optical network unit.
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 nonprovisional extension fee (37 CFR 1.17(a)) 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 mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to IMAD M KASSIM whose telephone number is (571)272-2958. The examiner can normally be reached 10:30AM-5:30PM, M-F (E.S.T.).
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, Michael J. Huntley can be reached at (303) 297 - 4307. 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.
/IMAD KASSIM/Primary Examiner, Art Unit 2129
Prosecution Insights