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 Amendment
The amendment filed on March 2nd, 2026 has been entered and Claim(s) 1-20 are -pending. Applicant’s amendments to the Claim(s) 17-20 have overcome the U.S.C 101 rejection directed to non-statutory subject matter previously set forth in the Non-Final Office Action mailed on December 1st, 2025.
Response to Arguments
Applicant’s arguments filed on March 2nd, 2026 have been fully considered but they are not persuasive.
In regards to U.S.C 101, applicant argues on page 11, that “claim 1 is not directed to a mental process.” This is not persuasive. Claim 1, as amended recites “select, based on a sample selection criteria, a subset of the samples from a first batch of the plurality of sequential batches as representative samples for the first batch,” recites evaluating data (i.e., samples) based on the data using a performance metric, and selecting a subset based on the evaluation, which therefore, describes a mental process (i.e., observation, evaluation, judgement). Although the claims recite a “wireless communication system,” this merely just identifies a field of use that is generically recited.
In regards to U.S.C 101, applicant argues on page 12, “claim 1 as a whole integrates into a practical application.” This is not persuasive. Claim 1 recites a wireless communication system and receiving samples associated with the wireless communication system and training a model to predict parameters. However, the claim does not recite any improvement to the operation of the wireless communication system. Rather the claim recites collecting data, evaluating the data, selecting samples, storing the selected samples, and training the model using the selected samples. The additional elements merely apply the abstract idea in a “wireless communication system,” which merely identifies a field of use that is generically recited.
In regards to applicants’ arguments on page 13, “claim 1 represents an improvement” This is not persuasive. Applicant refers to the specification that describes improvements associated with continual learning in wireless communication systems. However, the claim does not recite those improvements. However, the claim does not recite how improvement to the operation of the wireless communication system. Rather, the claim recites collecting data, evaluating the data, selecting samples, storing the selected samples, and training the model using the selected samples. See MPEP 2106.05(a), “After the examiner has consulted the specification and determined that the disclosed invention improves technology, the claim must be evaluated to ensure the claim itself reflects the disclosed improvement in technology. Intellectual Ventures I LLC v. Symantec Corp., 838 F.3d 1307, 1316, 120 USPQ2d 1353, 1359 (Fed. Cir. 2016) (patent owner argued that the claimed email filtering system improved technology by shrinking the protection gap and mooting the volume problem, but the court disagreed because the claims themselves did not have any limitations that addressed these issues). That is, the claim must include the components or steps of the invention that provide the improvement described in the specification.”
Applicant’s arguments with respect to claim(s) 1, 9, 17 have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument. Claim 1 recited a system performance metric which was generically recited and claim language did not require to be a system performance metric of wireless communication system, therefore, Pham does not need to disclose or suggest a system performance metric of wireless communication system.
Applicant’s arguments with respect to “as agreed during the interview, Butvinik in view of Pham does not disclose or suggest a wireless communication system”, referring to the interview summary of February 25th, 2026, Examiner noted that the proposed amendment would more clearly recite a wireless communication system performance metric and not generic system performance metric. Thus, in view of the new ground of rejection, the amended claims is viewed by Butvinik, as modified by Pham, further in view of Yuan.
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.
Claim(s) 1-3, 5-6, 9-11, 13-14, and 17-19 is/are rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (i.e., an abstract idea) without significantly more.
Regarding claim 1 and analogous claim 9 and 17:
Step 1 (whether a claim is to a statutory category):
Yes, the claim is within the four statutory categories (a process, machine, manufacture or composition of matter). Claim 1 recites a system, therefore, falls within a machine category. Claim 9 recites a method, therefore, falls within a process category. For claim 17, recites a non-transitory computer-readable storage medium, which falls within a manufacture category.
Step 2A Prong 1 (whether a claim is directed to a judicial exception):
Yes, “select, based on a sample selection criteria, a subset of the samples from a first batch of the plurality of batches as representative samples for the first batch,” describes a mental process (i.e., observation, evaluation, judgement) wherein observing a set of data one can evaluate and select a subset sample from the set of data by judging the rules/criteria for the selection process based on the performance (see MPEP 2106.04(a)(2)(III)).
Step 2A Prong 2 (evaluate whether the claim recites additional elements that integrate
the exception into a practical application):
No, “memory configured to store a model for predicting one or more parameters of the wireless communication system; and one or more hardware-based processors configured to:” describes an additional element as “apply it”, or merely including instructions to implement an abstract idea on a computer, or merely using a computer as a tool to perform an abstract idea (see MPEP 2106.05(f)). And, “wherein the sample selection criteria is based on a system performance metric, of the wireless communication system, computed for each of the samples;” is an additional element that generally links the use of the judicial exception to a particular technological environment or field of use. See MPEP §§ 2106.04(d), 2106.05(h). And “receive samples of the wireless communication system over a plurality of sequential batches, wherein each batch of the plurality of sequential batches represents a different, non-overlapping period of time; for each of the batches:” (These elements represent generic computer functions (i.e., mere data gathering in conjunction with the abstract idea)). And “store the subset of the samples for one or more batches of the plurality of sequential batches in the memory; and” describes additional elements that integrate the judicial exception into a practical application with the words "apply it" (or an equivalent), such as mere (i.e., selecting a particular data source or type of data to be manipulated) to implement an abstract idea on a computer (see MPEP 2106.05(f)). “upon receiving samples for a second batch, train the model to predict the one or more parameters using the samples from the second batch and the subset of the samples stored in the memory” describes additional elements that integrate the judicial exception into a practical application with the words "apply it" (or an equivalent), such as mere (i.e., selecting a particular data source or type of data to be manipulated) to implement an abstract idea on a computer (see MPEP 2106.05(f)).
Step 2B (Inventive concept):
No, it does not add significantly more since the intended practical application is well-understood, routine, and conventional and stated at a generic level (i.e. “apply it”, see MPEP 2106.05(f)).
Therefore, claim 1, 9 and 17 is/are ineligible.
Regarding claim 2 and analogous claim 10 and 18:
Further modifies the abstract idea of claim 1.
Step 2A Prong 2 (evaluate whether the claim recites additional elements that integrate
the exception into a practical application):
No, “wherein the sample selection criteria comprises samples in the first batch that have relatively low system performance compared to other samples in the first batch” is an additional element that amounts to adding insignificant extra-solution activity to the judicial exception. See MPEP 2106.04(d), 2106.05(g).
Step 2B (Inventive concept):
No, the limitation is an additional element that amounts to adding insignificant extra-solution activity to the judicial exception. See MPEP § 2106.05(g). The additional element is well‐understood, routine, and conventional See MPEP § 2106.05(d).
Therefore, claim 2, 10 and 18 is/are ineligible.
Regarding claim 3 and analogous claim 11 and 19:
Further modifies the abstract idea of claim 1.
Step 2A Prong 2 (evaluate whether the claim recites additional elements that integrate
the exception into a practical application):
No, “wherein the sample selection criteria is based on a bilevel optimization formulation that selects the subset of samples for storage within the memory” describes an additional element as “apply it”, or merely including instructions to implement an abstract idea on a computer, or merely using a computer as a tool to perform an abstract idea (see MPEP 2106.05(f)) and amounts to adding insignificant extra-solution activity to the judicial exception. See MPEP 2106.04(d), 2106.05(g).
Step 2B (Inventive concept):
No, it does not add significantly more since the intended practical application is well-understood, routine, and conventional and stated at a generic level (i.e. “apply it”, see MPEP 2106.05(f)) and the limitation is an additional element that amounts to adding insignificant extra-solution activity to the judicial exception. See MPEP § 2106.05(g). The additional element is well‐understood, routine, and conventional See MPEP § 2106.05(d).
Therefore, claim 3, 11 and 19 is/are ineligible.
Regarding claim 5 and analogous claim 13:
Further modifies the abstract idea of claim 1.
Step 2A Prong 2 (evaluate whether the claim recites additional elements that integrate
the exception into a practical application):
No, “wherein the one or more parameters comprise an estimate of channel state information for a wireless channel of the wireless communication system” is an additional element that amounts to adding insignificant extra-solution activity to the judicial exception. See MPEP 2106.04(d), 2106.05(g).
Step 2B (Inventive concept):
No, the limitation is an additional element that amounts to adding insignificant extra-solution activity to the judicial exception. See MPEP § 2106.05(g). The additional element is well‐understood, routine, and conventional See MPEP § 2106.05(d).
Therefore, claim 5 and 13 is/are ineligible.
Regarding claim 6 and analogous claim 14:
Further modifies the abstract idea of claim 1.
Step 2A Prong 2 (evaluate whether the claim recites additional elements that integrate
the exception into a practical application):
No, “wherein the one or more hardware-processors are configured to allocate resources within the wireless communication system is based on the predicted one or more parameters” describes an additional element as “apply it”, or merely including instructions to implement an abstract idea on a computer, or merely using a computer as a tool to perform an abstract idea (see MPEP 2106.05(f).)
Step 2B (Inventive concept):
No, it does not add significantly more since the intended practical application is well-understood, routine, and conventional and stated at a generic level (i.e. “apply it”, see MPEP 2106.05(f)).
Therefore, claim 6 and 14 is/are ineligible.
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 (i.e., changing from AIA to pre-AIA ) 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.
Claim(s) 1-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Butvinik et al., (US20220261633A1) in view of Pham et al., Non-Patent Literature “BILEVEL CONTINUAL LEARNING” further in view of Yuan et al., Non-Patent Literature “Transfer Learning and Meta Learning Based Fast Downlink Beamforming Adaptation.”
Regarding claim 1 and analogous claim 9 and 17:
Butvinik teaches:
A wireless communication system comprising ([0115], “Embodiments of the invention may fetch data (e.g., using data integration 402 of FIG. 4, initial fetch 502 of FIG. 5, and/or second fetch 602 of FIG. 6) over one or more communication channels, including a phone channel fetching example phone data shown in FIG. 7 (i.e., wherein wireless communication system)”)
memory configured to store a model for predicting one or more parameters of the wireless communication system; and ([0115], “Embodiments of the invention may fetch data (e.g., using data integration 402 of FIG. 4, initial fetch 502 of FIG. 5, and/or second fetch 602 of FIG. 6) over one or more communication channels, including a phone channel fetching example phone data shown in FIG. 7 (i.e., wherein wireless communication system)”…[0149], “an exemplary system for training a machine learning model using incremental learning without forgetting that may be used with embodiments of the present invention. Computing device 100 may include a controller or computer processor 105 that may be, for example, a central processing unit processor (CPU), a chip or any suitable computing device, an operating system 115, a memory ”…[0150], “Memory 120 may store for example, instructions (e.g. code 125)”
one or more hardware-based processors configured to: ([0149], “an exemplary system for training a machine learning model using incremental learning without forgetting that may be used with embodiments of the present invention. Computing device 100 may include a controller or computer processor 105 that may be, for example, a central processing unit processor (CPU), a chip or any suitable computing device, an operating system 115, a memory ”…[0150], “Memory 120 may store for example, instructions (e.g. code 125)”)
receive samples of the wireless communication system over a plurality of sequential batches, wherein each batch of the plurality of sequential batches represents a different, non-overlapping period of time; ([0115], “Embodiments of the invention may fetch data (e.g., using data integration 402 of FIG. 4, initial fetch 502 of FIG. 5, and/or second fetch 602 of FIG. 6) over one or more communication channels, including a phone channel fetching example phone data shown in FIG. 7 (i.e., wherein wireless communication)”…[0025], “receive or generate an ordered sequence of the plurality of training tasks and may incrementally train the model with each task sequentially in order (not in parallel) [sequential batches] (i.e., wherein ‘tasks’ is interpreted as batches, hence, not in parallel ‘non-overlapping period of time’). Each training task may be associated with one or more training samples [receive samples] and one or more corresponding labels respectively associated with the one or more training samples.”)
Butvinik does not explicitly teach:
for each batch of the plurality of sequential batches: select, based on a sample selection criteria, a subset of the samples from a first batch of the plurality of batches as representative samples for the first batch,
wherein the sample selection criteria is based on a system performance metric, of the wireless communication system, computed for each of the samples; store the subset of the samples for one or more batches of the plurality of sequential batches in the memory; and
upon receiving samples for a second batch of the plurality of sequential batches, train the model to predict the one or more parameters using the samples from the second batch and the subset of the samples stored in the memory.
Pham teaches:
for each batch of the plurality of sequential batches: (Section 2, “For each incoming mini batch of data B t n of task Tt, BCL initializes a fast weight φ to acquire the new knowledge in B t n .”)
select, based on a sample selection criteria, (Page 12, Section B, “To demonstrate this property, we consider the online class-incremental learning protocol ? in which the task identifier is not given to the model and it has to make predictions on all observed classes so far and consider the state-of-the-art method: Maximally Interfered Retrieval (MIR) (i.e., wherein under the broadest reasonable interpretation MIR is used for selecting). Instead of randomly sample a mini batch of data from the memory at each step, MIR works by selecting the replay data that maximize the model’s forgetting (i.e., wherein the use of MIR to target and select the replay data)”)
a subset of the samples from a first batch of the plurality of sequential batches as representative samples for the first batch, (Page 12, Section B, “We compare BCL-FO with and without MIR sampling strategies with the baselines in ? (i.e., wherein subset of samples, hence, baselines ‘representative samples’) and report the accuracy at the end of learning (ACC), forgetting measure (FM) in Table 4. We observe that BCL-FO-RAND consistently outperforms other methods with random sampling strategies and even comes close to ER-MIR which uses MIR sampling when 50 memory slots per class are allowed. When we replace random sampling in BCL-FO with MIR sampling, BCl-FO-MIR outperforms all the methods considered in both memory sizes, including ER-MIR.”)
store the subset of the samples for one or more batches of the plurality of sequential batches in the memory; and (Section 2, “When a new training sample arrives, BCL initializes a fast weight φ from the main model θ to learn this sample through experience replay with the episodic memory. Then, the trained fast weight is used to update the main model θ such that is can perform well on the generalization set. Fig. 1 illustrates our proposed BCL framework (i.e., wherein the subset samples are stored in memory)”)
upon receiving samples for a second batch of the plurality of sequential batches, (Algorithm 1, “Init: θ1, M ← ∅ Require: Memory management strategy for M for t ← 1 to T do Observe the dataset Dtr t sequentially for n ← 1 to n batches do Receive a mini batch of data Bn from Dtr (i.e., wherein the variable n can be 2, hence, ‘a second batch’ )”)
train the model to predict the one or more parameters using the samples from the second batch and the subset of the samples stored in the memory (Section 2.3, “During the inner optimization, current task data (i.e., wherein current task data is interpreted as the second batch) is mixed with previous data in the episodic memory (i.e., wherein samples stored in memory) for experience replay training [train the model]. However, previous data in the episodic memory are limited, which creates a bias towards the current task, which has more training data. Such bias will drive the model towards the current task, resulting in a performance degrade. To reduce this bias, we propose to regularize the inner optimization by preventing the fast-weight φ [one or more parameters] from deviating too much from the previous main models {θ}”)
Pham and Butvinik are both related to the same field of endeavor (i.e., machine learning). In view of the teachings of Pham it would have been obvious for a person of ordinary skill in the art to apply the teachings of Pham to Butvinik before the effective filing date of the claim invention in order to improve the efficiency of training a model using subset of samples over sequential batches in order to not forget learned tasks (Pham, Introduction, “Continual learning systems are specifically designed to learn continuously from a stream of tasks. They are able to accumulate knowledge over time to improve the future learning outcome, while still being able to perform well on the previous tasks. In the literature, prior works mainly focus on the continual learning protocol where the whole task data arrives at each step and the learner is allowed to train the current task on many epochs. This does not well reflect the real-world scenarios where data arrives sequentially and the learner has to learn new tasks on the fly. In this work, we make a next step towards the more realistic continual learning by developing our methods in the online continual learning regime where the training of each task is also performed in an online fashion with data arrives sequentially”)
Yuan teaches:
wherein the sample selection criteria is based on a system performance metric, of the wireless communication system, computed for each of the samples; (Page 21, para 2, “Based on the 20 fine-tuning samples, (i.e., wherein computed for each sample is interpreted as the 20 samples) we demonstrate the adaptation capability of the proposed algorithms via the SINR performance [system performance metric] (i.e., wherein SINR is signal-to-interference-plus-noise ratio which under the broadest reasonable interpretation is a system metric used in wireless communication to evaluate signal quality) using two different metrics in Fig. 4. Fig. 4(a) shows the effects of the transmit power on the SINR performance. As can be seen, the SINR increases as the transmit power increases for all schemes. The SINR result generated by the proposed meta-learning algorithm is very close to that of the BNN scheme which validates its effectiveness”)
Yuan and Butvinik are both related to the same field of endeavor (i.e., machine learning). In view of the teachings of Yuan it would have been obvious for a person of ordinary skill in the art to apply the teachings of Yuan to Butvinik before the effective filing date of the claim invention in order to improve the efficiency of wireless communication systems by using wireless communication system performance metrics (Yuan, Abstract, “This paper studies fast adaptive beamforming optimization for the signal-to-interference-plus-noise ratio balancing problem in a multiuser multiple-input single-output downlink system. Existing deep learning based approaches to predict beamforming rely on the assumption that the training and testing channels follow the same distribution which may not hold in practice. As a result, a trained model may lead to performance deterioration when the testing network environment changes. To deal with this task mismatch issue, we propose two offline adaptive algorithms based on deep transfer learning and meta-learning, which are able to achieve fast adaptation with the limited new labelled data when the testing wireless environment changes.”)
Regarding claim 2 and analogous claim 10 and 18:
Butvinik, as modified by Pham and Yuan, teaches the system of claim 1.
Butvinik further teaches:
samples in the first batch that have relatively low system performance compared to other samples in the first batch ([0026], “allows the replay samples to be used without labels (training data requires labels). Because labels are generated based on the model, and early model training has poor accuracy, early task iteration samples often have inaccurate labelling (i.e., wherein inaccurate labelling is interpreted as ‘low system performance’). Eliminating labels in the replay samples thus reduces inaccurate training based on those inaccurate labels to increase training accuracy compared to simply adding the replay samples and their labels to the training dataset. In addition, because the replay samples are used without labels, embodiments of the invention may use an aggregated distribution of replay samples zm (e.g., averages, standard deviations, modes, histograms, etc.), with which no labels are associated, rather than the discrete samples themselves. Training based on an aggregated distribution of replay samples, rather than the discrete replay samples themselves, provides a more distributed general knowledge of past training, without fixing the model to the exact past knowledge that often results in inflexible over-fitting that is less adaptable to training future tasks.”)
Butvinik does not explicitly teach:
wherein the sample selection criteria comprises
Pham further teaches:
wherein the sample selection criteria comprises (Page 12, Section B, “To demonstrate this property, we consider the online class-incremental learning protocol ? in which the task identifier is not given to the model and it has to make predictions on all observed classes so far and consider the state-of-the-art method: Maximally Interfered Retrieval (MIR) (i.e., wherein MIR is interpreted as a form of data sampling, hence, ‘sample selection criteria’). Instead of randomly sample a mini batch of data from the memory at each step, MIR works by selecting the replay data that maximize the model’s forgetting”)
The motivation for claim 2 is the same motivation for claim 1.
Regarding claim 3 and analogous claim 11 and 19:
Butvinik, as modified by Pham and Yuan, teaches the system of claim 1.
Butvinik does not explicitly teach:
wherein the sample selection criteria is based on a bilevel optimization formulation that selects the subset of samples for storage within the memory
Pham further teaches:
wherein the sample selection criteria is based on a bilevel optimization formulation that selects the subset of samples for storage within the memory (Page 2, paragraph 1, “Importantly, the generalization memory is never used to directly train the main model but only for improving its generalization. BCL learns new samples by first initializes a fast-weight and train it with experience replay using the episodic memory. Then, the trained fast-weight is used to update the original model such that it can generalize to the generalization memory [storage within the memory]. Therefore, BCL uses a bilevel optimization objective [bilevel optimization formulation] (Colson et al., 2007) with the inner problem as experience replay with the current data and the outer problem as optimizing the model’s performance on the generalization memory.”)
The motivation for claim 3 is the same motivation for claim 1.
Regarding claim 4 and analogous claim 12 and 20:
Butvinik, as modified by Pham and Yuan, teaches the system of claim 1.
Butvinik further teaches:
the one or more hardware-processors are configured to: ([0149], “an exemplary system for training a machine learning model using incremental learning without forgetting that may be used with embodiments of the present invention. Computing device 100 may include a controller or computer processor 105 that may be, for example, a central processing unit processor (CPU), a chip or any suitable computing device, an operating system 115, a memory ”…[0150], “Memory 120 may store for example, instructions (e.g. code 125)”)
retrain a model of the first batch using a sample pool of the first batch; ([0033], “Each task Ti is represented by Ti={xij, yij|j ϵ (1, . . . , Ni)}, where xij is the j-th example/sample of Ti, [first batch] (i.e., wherein each task is interpreted as the first task ‘batch’) and yij is its associated label. (xt, yt) denotes a test instance/example 200 [sample pool]. To simplify the notation, (xi, yi) denotes a training example from task Ti, omitting the second subscript j.”)
apply the retrained model to the sample pool to generate updated sample pool; ([0032], “for each test instance 200 adapt a classification network S for each test instance in order to classify the test instance. Since propagator g(·) 204 and the shared parameters φ0 will change during training for each new task [updated sample pool], forgetting can occur for previous tasks. To solve this problem, in training propagator g(·) 204 and classification network S for each task Ti, in addition to the training data of Ti, a small number of replayed samples may be generated by a data generation network”)
determine samples from updated sample pool based on the selection criteria; ([0059], “5: sample minibatch from T1 (i.e., wherein sample minibatch is interpreted as updated sample pool) //training DG 6: Minimize ψwae and update DS; //ψwae Wasserstein auto-encoder //training DPP (dynamic parameter propagator) add C //pn an Cn are batch of generated parameters and adapted computers respectively 7: Compute pn using eq.1 - eq.6 8: Construct Cn using pn and φ0 9: Minimize ψce and update g(·) and C end”)
determine whether the determined samples are same as samples determined in a previous iteration; and ([0026], “Using replay samples to constrain the model, (i.e., wherein determine samples) instead of simply inputting them into the training dataset, allows the replay samples to be used without labels (training data requires labels). Because labels are generated based on the model, and early model training has poor accuracy, early task iteration samples often have inaccurate labelling. Eliminating labels in the replay samples thus reduces inaccurate training based on those inaccurate labels to increase training accuracy compared to simply adding the replay samples and their labels to the training dataset (i.e., wherein determining accurate samples against prior iterations for accuracy)”)
repeat, as another iteration, retraining, applying, determining, and determining whether the determined samples are same as samples determined in the previous iteration until the determined samples are same as samples determined in the previous iteration, wherein the subset of samples comprise the determined samples. ([0026], “Using replay samples to constrain the model, (i.e., wherein determine samples) instead of simply inputting them into the training dataset, allows the replay samples to be used without labels (training data requires labels). Because labels are generated based on the model, and early model training has poor accuracy, early task iteration samples often have inaccurate labelling. Eliminating labels in the replay samples thus reduces inaccurate training based on those inaccurate labels to increase training accuracy compared to simply adding the replay samples and their labels to the training dataset (i.e., wherein determining accurate samples against prior iterations for accuracy)”… “A process or processor may train the machine learning model in a sequence of a plurality of sequential training iterations [repeat, as another iteration] respectively associated with the sequence of a plurality of training tasks. In each of the plurality of sequential training iterations the machine learning model is trained by iterating over operations 1120-1130 (i.e., wherein retraining is interpreted part of the process of the sequential training iterations). In operation 1120, a processor (e.g., controller 105 of FIG. 1) may generate the task-specific parameters for the current training iteration by applying a propagator [applying] to the one or more training samples associated with the current training task. The training of the model for the current training task may be constrained by one or more of the training samples associated with a previous training task in a previous training iteration. The model may be constrained to reduce minimize variations of one or more layer outputs of the model caused by changes in the subset of shared parameters and the propagator resulting from the current training iteration by using the one or more training samples associated with the previous training task. The propagator for the current training task may be generated based on the one or more of the training samples associated with the previous training task but not the corresponding labels respectively associated therewith. In some embodiments, the propagator may not be applied to the one or more training samples associated with the previous training task to generate the task-specific parameters for the current training task. The one or more of the training samples associated with the previous training task may be generated based on an aggregated distribution of a plurality of the training samples to which the propagator was applied in the previous training iteration.”)
Butnivik does not explicitly teach:
wherein to select the subset of samples
Pham further teaches:
wherein to select the subset of samples (Page 12, Section B, “We compare BCL-FO with and without MIR sampling strategies with the baselines in ? (i.e., wherein subset of samples, hence, baselines ‘representative samples’) and report the accuracy at the end of learning (ACC), forgetting measure (FM) in Table 4. We observe that BCL-FO-RAND consistently outperforms other methods with random sampling strategies and even comes close to ER-MIR which uses MIR sampling when 50 memory slots per class are allowed. When we replace random sampling in BCL-FO with MIR sampling, BCl-FO-MIR outperforms all the methods considered in both memory sizes, including ER-MIR.”)
The motivation for claim 4 is the same motivation for claim 1.
Regarding claim 5 and analogous claim 13:
Butvinik, as modified by Pham and Yuan, teaches the system of claim 1.
Butvinik, as modifies by Pham, does not explicitly teach:
wherein the one or more parameters comprise an estimate of channel state information for a wireless channel of the wireless communication system.
Yuan teaches:
wherein the one or more parameters comprise an estimate of channel state information for a wireless channel of the wireless communication system (Section V, paragraph 2, “the BS can obtain perfect CSI based on channel estimation or feedback. Each sample pair in all datasets is composed of channel realization and the uplink power allocation vector. For each channel realization, we can generate its associated uplink power allocation vector by solving the uplink problem (3). Channel realizations of the testing dataset and the adaption dataset come from the same distribution (i.e., wherein channel estimation is interpreted to include parameters (i.e., wireless signal), hence, for a wireless communication system)”)
The motivation for claim 5 is the same motivation for claim 1.
Regarding claim 6 and analogous claim 14:
Butvinik, as modified by Pham and Yuan, teaches the system of claim 1.
Butvinik, as modifies by Pham, does not explicitly teach:
wherein the one or more hardware-processors are configured to allocate resources within the wireless communication system is based on the predicted one or more parameters
Yuan further teaches:
wherein the one or more hardware-processors are configured to allocate resources within the wireless communication system is based on the predicted one or more parameters (Introduction, paragraph 1, Beamforming is recognized as one of the most promising multi-antenna techniques since it can efficiently improve the antenna diversity gain and mitigate multiuser interference (i.e., wherein allocate resources such as (i.e., multiuser interference))”…“improve the spectral efficiency of modern wireless communications systems due to their ability to exploit spatial characteristics of the propagation channel”…“This is the result of the mapping from the input channel state to output beamforming that is obtained by training the neural networks [predicted one or more parameters] (i.e., wherein the channel state is interpreted as the parameter predicted (i.e., how signals travel)) in an offline manner. The beamforming solution can be directly predicted using the trained network in real time. The advantage of the learning to optimize framework is to transfer the complex real-time optimization procedures to offline training showing great potential to solve the beamforming design problems in multi-antenna systems”))
The motivation for claim 6 is the same motivation for claim 1.
Regarding claim 7 and analogous claim 16:
Butvinik, as modified by Pham and Yuan, teaches the system of claim 6.
Butvinik, as modifies by Pham, does not explicitly teach:
wherein to allocate resources, the one or more hardware-processors are configured to control an allocation of power to a plurality of base stations of the wireless communication system for a given geographic region based on the predicted one or more parameters.
Yuan further teaches:
wherein to allocate resources, the one or more hardware-processors are configured to control an allocation of power to a plurality of base stations of the wireless communication system for a given geographic region based on the predicted one or more parameters (Section V, paragraph 2, “the BS can obtain perfect CSI based on channel estimation or feedback. Each sample pair in all datasets is composed of channel realization and the uplink power allocation vector. For each channel realization, we can generate its associated uplink power allocation vector [allocation of power]”…Section II, “A multi-input single-output (MISO) downlink transmission system is considered, in which K single antenna users are served by a base station (BS) [base stations] with M antennas (i.e., wherein allocate resources (antennas))”…Introduction, paragraph 2, “the training and testing channel data are drawn from the same distribution in a fixed stationary environment (i.e., wherein a fixed stationary environment is interpreted as ‘a given geographic region’)”).
The motivation for claim 7 is the same motivation for claim 1.
Regarding claim 8 and analogous claim 15:
Butvinik, as modified by Pham and Yuan, teaches the system of claim 6.
Butvinik, as modifies by Pham, does not explicitly teach:
wherein to allocate resources, the one or more hardware-processors are configured to control beamforming for a plurality of antennas of the wireless communication system based on the predicted one or more parameters.
Yuan further teaches:
wherein to allocate resources, the one or more hardware-processors are configured to control beamforming for a plurality of antennas of the wireless communication system based on the predicted one or more parameters (Introduction, paragraph 1“Multi-antenna techniques [plurality of antennas] have been widely used to improve the spectral efficiency of modern wireless communications systems due to their ability to exploit spatial characteristics of the propagation channel [1], [2]. Beamforming [beamforming] is recognized as one of the most promising multi-antenna techniques since it can efficiently improve the antenna diversity gain and mitigate multiuser interference (i.e., wherein allocate resources such as (i.e., multiuser interference))”…“This is the result of the mapping from the input channel state to output beamforming that is obtained by training the neural networks [predicted one or more parameters] (i.e., wherein the channel state is interpreted as the parameter predicted (i.e., how signals travel)) in an offline manner. The beamforming solution can be directly predicted using the trained network in real time. The advantage of the learning to optimize framework is to transfer the complex real-time optimization procedures to offline training showing great potential to solve the beamforming design problems in multi-antenna systems”)
The motivation for claim 8 is the same motivation for claim 1.
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 AMINA BENOURAIDA whose telephone number is (571)272-4340. The examiner can normally be reached Monday-Friday 8:30am-5pm ET..
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.
/AMINA MORENO BENOURAIDA/Examiner, Art Unit 2129
/MICHAEL J HUNTLEY/Supervisory Patent Examiner, Art Unit 2129
Prosecution Insights