DETAILED ACTION
This nonfinal office action is responsive to the claims filed on March 6, 2024. Claims 81-99 are pending. Claims 81, 90, and 95 are independent.
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 .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on May 21, 2024 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
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.
Claims 81--89 are rejected under 35 U.S.C. 103 as being unpatentable over Feriani et al. (Single and Multi-Agent Deep Reinforcement Learning for AI-Enabled Wireless Networks: A Tutorial), hereinafter Feriani, in view of Wilhelmi et al. (Usage of Network Simulators in Machine-Learning-Assisted 5G/6G Networks), hereinafter Wilhelmi.
Wilhelmi was cited in Applicant’s IDS dated 5/21/2024. Examiner has relied upon an earlier version of Wilhelmi which is being provided.
Regarding claim 81, Feriani teaches:
at least one processor, (Feriani, page 2, column 1, paragraph 1: “However, the training phase of DNNs requires a considerable amount of computation power which necessitates the use of GPUs and high-performance CPU clusters.”)
deriving, based on quality requirements in relation to reinforcement learning, a preliminary reinforcement learning plan, (Feriani, page 13, column 2, paragraph 3: “Planning is therefore performed by applying the known RL methods such as Q-learning to estimate value functions using the simulated and real experience. The estimated value functions are thereby utilized to optimize or improve a policy.” – The optimization or improvement of a policy is analogous to a preliminary reinforcement learning plan which is being based on the value functions, e.g., the quality requirements.)
revising, based on data related to reinforcement learning on a network scenario, said preliminary reinforcement learning plan to a final reinforcement learning plan, and (Feriani, page 13, column 2, paragraph 3: “The planning is performed by selecting uniformly random K initial state-action pairs (s,a), simulating the environment to obtain the next states and rewards, and update the Q-values of the sampled pairs.” – Updating the Q-values of the sampled pairs based off the simulated environment is analogous to revising the plan based on data related to reinforcement learning on a network scenario as Feriani teaches the RL in communication networks – see abstract.)
Feriani does not explicitly teach:
at least one memory including computer program code, and at least one interface configured for communication with at least another apparatus, the at least one processor, with the at least one memory and the computer program code, being configured to cause the apparatus to perform:
transmitting said final reinforcement learning plan to an artificial intelligence pipeline orchestrating entity.
However, Wilhelmi teaches:
at least one memory including computer program code, and at least one interface configured for communication with at least another apparatus, the at least one processor, with the at least one memory and the computer program code, being configured to cause the apparatus to perform: (Wilhelmi, page 162, paragraph 2: “The ITU ML architecture defines a set of logical components, interfaces, and procedures to realize ML-assisted communications.”)
transmitting said final reinforcement learning plan to an artificial intelligence pipeline orchestrating entity. (Wilhelmi, Fig. 1: Steps 3 and 4 are transmitting the model to the marketplace which further sends it to the ML Underlay Network in the pipeline which is analogous to transmitting the model configuration to an artificial intelligence pipeline orchestrating entity.)
PNG
media_image1.png
638
1108
media_image1.png
Greyscale
Wilhelmi is considered analogous to the claimed invention as it is in the same field of endeavor, machine learning in communications environments. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to have modified Feriani, which already teaches an apparatus to derive and revise a reinforcement learning plan but does not explicitly teach transmitting the plan to a pipeline orchestrating entity, to include the teachings of Wilhelmi which does teach transmitting the plan to a pipeline orchestrating entity in order to provide interoperability which is "meant to enable a seamless integration of intelligent network functionalities in the communication network.” (Wilhelmi, page 164, column 1, paragraph 1)
Regarding claim 82, Feriani and Wilhelmi teach the apparatus of claim 81, as cited above.
Feriani further teaches:
generating, based on said final reinforcement learning plan, a reinforcement learning configuration. (Feriani, Fig. 5(a) below – The plan and policy are sent to the environment, where actions are taken and then the model is further updated, e.g. the configuration is generated based on the plan.)
PNG
media_image2.png
604
596
media_image2.png
Greyscale
Regarding claim 83, Feriani and Wilhelmi teach the apparatus of claim 82, as cited above.
Feriani does not explicitly teach:
said reinforcement learning configuration includes a reinforcement learning monitoring configuration comprising at least one of information on parameters to be monitored, information on parameters to be reported, and information on a measurement period.
However, Wilhelmi further teaches:
said reinforcement learning configuration includes a reinforcement learning monitoring configuration comprising at least one of information on parameters to be monitored, information on parameters to be reported, and information on a measurement period. (Wilhelmi, page 164, column 1, paragraph 1: “In particular, the features that may facilitate the interoperability of out-of-the-box simulators are the support for Command-Line Interface (CLI) execution mode, the level of monitoring supported (real-time, batch, model-based, and so on), automation of data collection, and in applying the ML output in the simulator (e.g., reading from log fi les vs. API-based interface with ML functions).” – The level of monitoring supported is analogous to the configurations including information on parameters to be monitored or reported.)
Regarding claim 84, Feriani and Wilhelmi teach the apparatus of claim 82, as cited above.
Feriani further teaches:
said reinforcement learning configuration includes a reinforcement learning trustworthiness configuration comprising at least one of a reinforcement learning model explainability configuration, a reinforcement learning model fairness configuration, and a reinforcement learning model robustness configuration. (Feriani, page 13, column 1, paragraph 2: “A common taxonomy classifies interpretability models along two main dimensions: the scope or level of the explanans (global vs local) and the time when the explanations are generated (intrinsic vs post-hoc). Global approaches explain the behavior of the whole model, whereas local ones provide explanations to local predictions. The intrinsic or transparent category encompasses models constructed to be self-explanatory by reducing their complexity.” – The interpretability is analogous to the trustworthiness including model explainability.)
Feriani does not explicitly teach that the trustworthiness is included in the configurations, however Wilhelmi teaches this:
(Wilhelmi, page 161, column 2, paragraph 3: “Furthermore, sRL aims to minimize the negative effects that unconstrained exploration methods can produce during the learning procedure. This can be achieved either by adding extra information to the exploration mechanism (e.g., external advice) or by applying certain risk-aware criteria (e.g., exploration based on water-filling methods).”
Regarding claim 85, Feriani and Wilhelmi teach the apparatus of claim 82, as cited above.
Feriani does not explicitly teach:
providing said reinforcement learning configuration to an artificial intelligence trust management entity, and providing said final reinforcement learning plan to said artificial intelligence trust management entity.
However, Wilhelmi further teaches:
providing said reinforcement learning configuration to an artificial intelligence trust management entity, and providing said final reinforcement learning plan to said artificial intelligence trust management entity. (Wilhelmi, Fig. 1 – “Management Subsystem” contains the policies and constraints of the model and therefore is being provided the reinforcement learning configuration and acting as an artificial intelligence trust management entity.)
Regarding claim 86, Feriani and Wilhelmi teach the apparatus of claim 85, as cited above.
Feriani does not explicitly teach:
receiving, from said artificial intelligence trust management entity, metrics in accordance with said reinforcement learning configuration collected by at least one of an artificial intelligence data source management entity, an artificial intelligence training management entity, and an artificial intelligence inference management entity.
However, Wilhelmi further teaches:
receiving, from said artificial intelligence trust management entity, metrics in accordance with said reinforcement learning configuration collected by at least one of an artificial intelligence data source management entity, an artificial intelligence training management entity, and an artificial intelligence inference management entity. (Wilhelmi, Fig. 1 – “ML Marketplace” is receiving the model, which includes the metrics in accordance with the configuration, from the Management subsystem which is analogous to the artificial intelligence data source management entity receiving the metrics from the trust management entity.)
Regarding claim 87, Feriani and Wilhelmi teach the apparatus of claim 82, as cited above.
Feriani does not explicitly teach:
transmitting said reinforcement learning configuration to at least one of an artificial intelligence data source management entity, an artificial intelligence training management entity, and an artificial intelligence inference management entity, and
receiving metrics in accordance with said reinforcement learning configuration from at least one of said artificial intelligence data source management entity, said artificial intelligence training management entity, and said artificial intelligence inference management entity.
However, Wilhelmi further teaches:
transmitting said reinforcement learning configuration to at least one of an artificial intelligence data source management entity, an artificial intelligence training management entity, and an artificial intelligence inference management entity, and (Wilhelmi, Fig. 1 – Select Model and Update model being sent from Management subsystem to ML Marketplace would include transmitting the configuration.)
receiving metrics in accordance with said reinforcement learning configuration from at least one of said artificial intelligence data source management entity, said artificial intelligence training management entity, and said artificial intelligence inference management entity. (Wilhelmi, Fig. 1 – The model monitoring being sent to the Management Subsystem would include the metrics in accordance with said reinforcement learning configuration.)
Regarding claim 88, Feriani and Wilhelmi teach the apparatus of claim 86, as cited above.
Feriani further teaches:
verifying a level of safety of said reinforcement learning based on said metrics. (Feriani, page 22, column 1, paragraph 3: “Constrained/safe RL: RL is based on maximizing the reward feedback. The reward function is designed by human experts to guide the agent policy search but reward design is often challenging and can lead to unintended behavior. Wireless communication problems are often formulated as optimization problems under constraints. To account for those constraints, most of the recent works adopt a reward shaping strategy where penalties are added to the reward function for violating the defined constraints. In addition, reward shaping does not ensure that the exploration during the training is constraint-satisfying. This motivates the constrained RL framework. It enables the development of more reliable algorithms ensuring that the learned policies satisfy reasonable service quality or/and respect system constraints.” – The constrained RL is optimizing the RL parameters under constraints for safe RL which is analogous to verifying the level of safety of said reinforcement learning based on the metrics.)
Regarding claim 89, Feriani and Wilhelmi teach the apparatus of claim 88, as cited above.
Feriani further teaches:
modifying said final reinforcement learning plan based on said level of safety of said reinforcement learning. (Feriani, page 22, column 1, paragraph 3: “In addition, reward shaping does not ensure that the exploration during the training is constraint-satisfying. This motivates the constrained RL framework. It enables the development of more reliable algorithms ensuring that the learned policies satisfy reasonable service quality or/and respect system constraints.” – The constrained RL framework is used to modify the parameters to ensure the safety constraints are satisfied.)
Claims 90-99 are is rejected under 35 U.S.C. 103 as being unpatentable Wilhelmi in view of Feriani.
Regarding claim 90, Wilhelmi teaches the apparatus:
at least one processor, at least one memory including computer program code, and at least one interface configured for communication with at least another apparatus, the at least one processor, with the at least one memory and the computer program code, being configured to cause the apparatus to perform: (Wilhelmi, page 162, paragraph 2: “The ITU ML architecture defines a set of logical components, interfaces, and procedures to realize ML-assisted communications.”)
receiving a machine learning configuration, receiving a final machine learning plan,
transmitting said machine learning configuration to at least one of an artificial intelligence data source management entity, an artificial intelligence training management entity, and an artificial intelligence inference management entity, and (Wilhelmi, Fig. 1 below – The definition of use cases, which the Management subsystem has already received, includes the policies and constraints which are analogous to the configuration and final machine learning plan. Steps 3 and 4 are transmitting the ML model in the pipeline which is analogous to transmitting the machine learning configuration to an artificial intelligence source management entity.)
receiving metrics in accordance with said machine learning configuration from at least one of said artificial intelligence data source management entity, said artificial intelligence training management entity, and said artificial intelligence inference management entity. (Wilhelmi, Fig. 1 – The model monitoring being sent to the Management Subsystem would include the metrics in accordance with said machine learning configuration.)
PNG
media_image1.png
638
1108
media_image1.png
Greyscale
Wilhelmi does not explicitly teach:
That the machine learning is reinforcement learning
However, Feriani teaches:
That the machine learning is reinforcement learning (Feriani, page 1, column 2, paragraph 2: “The recent success of AI techniques, namely Machine Learning (ML) and Deep Learning (DL), has spurred the adoption of a learning perspective to solve wireless control and management problems. For instance, Deep Neural Networks (DNN) are universal approximators able to estimate any function thus they can approximate optimal solutions for complex tasks. DNNs can be used in three different ML settings: supervised ML, unsupervised ML, and Reinforcement Learning (RL).”)
Feriani is considered analogous to the claimed invention as it is in the same field of endeavor, machine learning in communications networks. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to have modified Wilhelmi, which already teaches sending and receiving the machine learning plan, configuration and metrics to and from an artificial intelligence data source management entity but does not explicitly teach that the machine learning is reinforcement learning, to include the teachings of Feriani which does teach that the machine learning is reinforcement learning in order to "rely on “trial and error” to solve sequential decision-making problems." (Feriani, page 1, column 2, paragraph 2)
Regarding claim 91, Wilhelmi teaches the apparatus of claim 90, as cited above.
Wilhelmi further teaches:
said reinforcement learning configuration includes a reinforcement learning monitoring configuration comprising at least one of information on parameters to be monitored, information on parameters to be reported, and information on a measurement period. (Wilhelmi, page 164, column 1, paragraph 1: “In particular, the features that may facilitate the interoperability of out-of-the-box simulators are the support for Command-Line Interface (CLI) execution mode, the level of monitoring supported (real-time, batch, model-based, and so on), automation of data collection, and in applying the ML output in the simulator (e.g., reading from log fi les vs. API-based interface with ML functions).” – The level of monitoring supported is analogous to the configurations including information on parameters to be monitored or reported.)
Regarding claim 92, Wilhelmi teaches the apparatus of claim 90, as cited above.
said reinforcement learning configuration includes a reinforcement learning trustworthiness configuration … (Wilhelmi, page 161, column 2, paragraph 4: “Given the lack of general mechanisms and procedures for providing trustworthy ML-aware communications, we devise the potential usage of network simulators for training, testing, and evaluating the effect of ML models before being applied to operative networks. In particular, simulators can provide diverse functionalities to enhance the confidence level of future ML-assisted networks.”)
Wilhelmi does not explicitly teach:
… comprising at least one of a reinforcement learning model explainability configuration, a reinforcement learning model fairness configuration, and a reinforcement learning model robustness configuration.
However, Feriani teaches:
… comprising at least one of a reinforcement learning model explainability configuration, a reinforcement learning model fairness configuration, and a reinforcement learning model robustness configuration. (Feriani, page 13, column 1, paragraph 2: “A common taxonomy classifies interpretability models along two main dimensions: the scope or level of the explanans (global vs local) and the time when the explanations are generated (intrinsic vs post-hoc). Global approaches explain the behavior of the whole model, whereas local ones provide explanations to local predictions. The intrinsic or transparent category encompasses models constructed to be self-explanatory by reducing their complexity.” – The interpretability is analogous to the trustworthiness including model explainability.)
Regarding claim 93, Wilhelmi teaches the apparatus of claim 90, as cited above.
Wilhelmi further teaches:
said final reinforcement learning plan includes at least one of a list of actions allowed to be executed for reinforcement learning, information on an expected impact of application of said final reinforcement learning plan on a network corresponding to said network scenario, information on an expected time interval of said expected impact, and information on measures to be taken upon exceedance of said expected impact and/or said expected time interval of said expected impact. (Wilhelmi, page 163, column 2, paragraph 4: “In ns3-gym, the synchronization between the simulator and the ML components is achieved through a discretized mapping of the simulation’s information (e.g., channel status, number of nodes, traffic demands) with the states, actions, and rewards to be used by an ML algorithm. The whole procedure is carried out in execution time.” – The information with the states, actions and rewards to be used by a ML algorithm is analogous to the learning plan including a list of actions allowed to be executed for reinforcement learning.)
Regarding claim 94, Wilhelmi teaches the apparatus of claim 93, as cited above.
Wilhelmi does not explicitly teach:
each of said at least one action is defined by at least one of information on one or more parameters to be changed by said action, information on one or more allowable change ranges corresponding to said one or more parameters to be changed by said action, information on one or more action targets, information on an action execution time, information on an action execution frequency, information on an action application realm, information on an expected impact of application of said action on said network, information on an expected time interval of said expected impact, and information on measures to be taken upon exceedance of said expected impact and/or said expected time interval of said expected impact.
However, Feriani further teaches:
each of said at least one action is defined by at least one of information on one or more parameters to be changed by said action, information on one or more allowable change ranges corresponding to said one or more parameters to be changed by said action, information on one or more action targets, information on an action execution time, information on an action execution frequency, information on an action application realm, information on an expected impact of application of said action on said network, information on an expected time interval of said expected impact, and information on measures to be taken upon exceedance of said expected impact and/or said expected time interval of said expected impact. (Feriani, page 4, column 2, paragraph 3: “The full observability assumption of MDPs enables the agent to access the exact state of the system s at every time step t. Given the state s, the agent will decide to take an action a transiting the system to a new state s’ sampled from the probability distribution P(.|s,a).” – The action taken transiting the system to a new state s’ is analogous to the information on the expected impact of application of said action on said network.)
Regarding claim 95, Wilhelmi teaches the apparatus:
at least one processor, at least one memory including computer program code, and at least one interface configured for communication with at least another apparatus, the at least one processor, with the at least one memory and the computer program code, being configured to cause the apparatus to perform: (Wilhelmi, page 162, paragraph 2: “The ITU ML architecture defines a set of logical components, interfaces, and procedures to realize ML-assisted communications.”)
receiving a final machine learning plan, (Wilhelmi, Fig. 1 below – The definition of use cases, which the Management subsystem has already received, includes the constraints which are analogous to the final machine learning plan. Steps 3 and 4 are transmitting the ML model in the pipeline which is analogous to transmitting the machine learning configuration to an artificial intelligence source management entity.)
PNG
media_image1.png
638
1108
media_image1.png
Greyscale
deciding on a degree of implementation of said final machine learning plan, and implementing said final machine learning plan based on said decided degree of implementation. (Wilhelmi, page 163, column 1, paragraph 2: “In this regard, the management subsystem must deploy, configure, and interact with the simulated functionalities that are required by the ML use case. Notice that, given the diversity of simulation tools (stored and maintained in a repository), an interoperability plugin is required to translate simulator-specific commands into standardized operations.” – The ML use case determines the functionalities of the ML model which is analogous to a degree of implementation of the plan.)
Wilhelmi does not explicitly teach:
That the machine learning is reinforcement learning
However, Feriani teaches:
That the machine learning is reinforcement learning (Feriani, page 1, column 2, paragraph 2: “The recent success of AI techniques, namely Machine Learning (ML) and Deep Learning (DL), has spurred the adoption of a learning perspective to solve wireless control and management problems. For instance, Deep Neural Networks (DNN) are universal approximators able to estimate any function thus they can approximate optimal solutions for complex tasks. DNNs can be used in three different ML settings: supervised ML, unsupervised ML, and Reinforcement Learning (RL).”)
Feriani is considered analogous to the claimed invention as it is in the same field of endeavor, machine learning in communications networks. Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date to have modified Wilhelmi, which already teaches receiving a machine learning plan and deciding on a degree of implementation prior to implementing the machine learning plan but does not explicitly teach that the machine learning is reinforcement learning, to include the teachings of Feriani which does teach that the machine learning is reinforcement learning in order to "rely on “trial and error” to solve sequential decision-making problems." (Feriani, page 1, column 2, paragraph 2)
Regarding claim 96, Wilhelmi teaches the apparatus of claim 95, as cited above.
Wilhelmi further teaches:
said deciding is based on current network conditions. (Wilhelmi, page 164, column 1, paragraph 1: “Interoperability is, therefore, meant to enable a seamless integration of intelligent network functionalities in the communication network. For that, it is imperative that the simulated network functionalities are managed using the same operation and maintenance mechanisms as for the network functionalities in the ML underlay. In this regard, the interoperability plugin is crucial to handle the different simulated networks in execution time, thus allowing for standardized functionalities such as start or stop.” – The interoperability integrating the functionalities being crucial to the different networks in execution time is analogous to the deciding based on current network conditions.)
Regarding claim 97, Wilhelmi teaches the apparatus of claim 95, as cited above.
Wilhelmi further teaches:
transmitting information on said degree of implementation. (Wilhelmi, page 163, column 1, bullet 3: “The management subsystem selects the ML model from the marketplace, according to the use-case metadata, the optimization goals, and the available models.” – The use-case metadata is analogous to the degree of implementation.)
Regarding claim 98, Wilhelmi teaches the apparatus of claim 95, as cited above.
Wilhelmi further teaches:
said final reinforcement learning plan includes at least one of a list of actions allowed to be executed for reinforcement learning, information on an expected impact of application of said final reinforcement learning plan on a network corresponding to said network scenario, information on an expected time interval of said expected impact, and information on measures to be taken upon exceedance of said expected impact and/or said expected time interval of said expected impact. (Wilhelmi, page 163, column 2, paragraph 4: “In ns3-gym, the synchronization between the simulator and the ML components is achieved through a discretized mapping of the simulation’s information (e.g., channel status, number of nodes, traffic demands) with the states, actions, and rewards to be used by an ML algorithm. The whole procedure is carried out in execution time.” – The information with the states, actions and rewards to be used by a ML algorithm is analogous to the learning plan including a list of actions allowed to be executed for reinforcement learning.)
Regarding claim 99, Wilhelmi teaches the apparatus of claim 98, as cited above.
Wilhelmi does not explicitly teach:
said list of actions allowed to be executed for said reinforcement learning includes at least one action, wherein each of said at least one action is defined by at least one of information on one or more parameters to be changed by said action, information on one or more allowable change ranges corresponding to said one or more parameters to be changed by said action, information on one or more action targets, information on an action execution time, information on an action execution frequency, information on an action application realm, information on an expected impact of application of said action on said network, information on an expected time interval of said expected impact, and information on measures to be taken upon exceedance of said expected impact and/or said expected time interval of said expected impact.
However, Feriani further teaches:
said list of actions allowed to be executed for said reinforcement learning includes at least one action, wherein each of said at least one action is defined by at least one of information on one or more parameters to be changed by said action, information on one or more allowable change ranges corresponding to said one or more parameters to be changed by said action, information on one or more action targets, information on an action execution time, information on an action execution frequency, information on an action application realm, information on an expected impact of application of said action on said network, information on an expected time interval of said expected impact, and information on measures to be taken upon exceedance of said expected impact and/or said expected time interval of said expected impact. (Feriani, page 4, column 2, paragraph 3: “The full observability assumption of MDPs enables the agent to access the exact state of the system s at every time step t. Given the state s, the agent will decide to take an action a transiting the system to a new state s’ sampled from the probability distribution P(.|s,a).” – The action taken transiting the system to a new state s’ is analogous to the information on the expected impact of application of said action on said network.)
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Bhargava et al. (US20200005168)
Shi et al. (Reinforcement Learning for Dynamic Resource Optimization in 5G Radio Access Network Slicing)
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JACQUELINE MEYER whose telephone number is (703)756-5676. The examiner can normally be reached M-F 8:00 am - 4:30 pm EST.
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, Tamara Kyle can be reached at 571-272-4241. 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.
/J.C.M./Examiner, Art Unit 2144
/TAMARA T KYLE/Supervisory Patent Examiner, Art Unit 2144
Prosecution Insights