Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
DETAILED ACTION
This action is responsive to the following communication: Amendment filed Apr. 20, 2026. This Action is made Final.
Claims 1-3, 5-17 are pending in the case. Claims 1 and 12-15 are independent claims.
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 1-3, 5-14, 16 and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Nakayama et al. (hereinafter Nakayama) U.S. Patent Publication No. 2021/0406782 in view of Bhattacharjee et al. (hereinafter Bhatt) U.S. Patent Publication No. 2020/0334567.
With respect to independent claim 1, Nakayama teaches a prediction model training method, which is performed by a server, comprising:
transmitting, to a plurality of training devices, a model to be trained by the plurality of training devices (see e.g., Abstract Para [42][43] and Claim 1 – “ each of the aggregators sends the semi-global machine learning model to the associated agents” ”each of the aggregators sends the semi-global machine learning model to the associated agents, and each of the agents updates the local machine learning model with the semi-global machine learning model received from the associated aggregator.”), wherein the model to be trained comprises feature extraction layers configured to extract user features and prediction layers configured to perform information prediction (see e.g., Para [12]-[16] [21]-[29][122][123]- “Models trained in advance with prepared training data Comparisons only against the static base model Static models are trained for limited set of real world scenarios. Single model is deployed”);
classifying the plurality of training devices into at least one group based on the user features extracted by the training devices (see e.g., Para [32]-[40][145]- “multiple aggregators coupled to the communication network and each uniquely associated with the agents, each aggregator comprising a model collector collecting the local machine learning models from the associated agents”” The personalization can be extendedly interpreted as a model aggregation for each group of users who share a similar behavioral pattern. The group-level model management and preparation virtually cluster all the users into multiple groups by incorporating a feature vector-based clustering method. This enables the customization and advanced control of ML models distributed by aggregators for different types of users.”);
receiving, from the plurality of training devices, model parameters obtained by the respective training devices training the model to be trained (see e.g., Para [141]-[144), wherein the model parameters comprise first parameters corresponding to the feature extraction layers and second parameters corresponding to the prediction layers (see e.g., Para [21]-[29][119]-[123]- “The network comprises three sequentially connected blocks of layers. (a) The embedding block 80 takes the process model state as an input and converts it into a common representation by accounting for the heterogeneity of process models. The embedding block 80 additionally predicts process variables that are not part of the process state but can be descriptive of the process performance and other metrics that an operator cares about. Such variables are determined by each agent's operator. (b) The inference block 90, whose parameters are aggregated through the federated learning process, uses the common representation of the input to produce an output. Since the inference block 90 is agnostic to process model variations, it can be generated by aggregating inference blocks 90 across the network. (c) The transfer block 100 converts the common representation of the output into an output value understood by the particular process model. The transfer block 100 can also predict process variables aside from ones assigned to the embedding block 80. Process variables are used by inference and transform blocks 80, 90 to calculate the output.”);
performing global federated aggregation based on the first parameters to obtain a global federated aggregation result (see e.g., Fig. 10 Para [50][110][137]-[144]- “the system may further comprise a model repository storing the global machine learning models previously created by the system and meta-data indicating tasks used for training the respective global machine learning models”);
performing intra-group federated aggregation for each of the at least one group, based on the second parameters of one or more of the plurality of training devices in a respective group, among each of the at least one group, to obtain an intra-group federated aggregation result (see e.g., Para [32]-[40][145] [144]- “this two model approach, for each agent we first randomly initialize the two models”); and
transmitting, to each of the plurality of training devices, the global federated aggregation result and the intra-group federated aggregation result associated the respective group of the respective training device (see e.g., Para [32]-[40][145] [144]), so that the plurality of training devices update the first parameters of the feature extraction layers based on the global federated aggregation result and update the second parameters of the prediction layers based on the intra-group federated aggregation result (see e.g., Para [141]-[144]- “he personalization can be extendedly interpreted as a model aggregation for each group of users who share a similar behavioral pattern. The group-level model management and preparation virtually cluster all the users into multiple groups by incorporating a feature vector-based clustering method. This enables the customization and advanced control of ML models distributed by aggregators for different types of users.”).
Nakayama does not expressly show wherein the classifying the plurality of training devices into the at least one group comprises: acquiring process capabilities of each of the plurality of training devices; and classifying the plurality of training devices into one or more groups among the at least one group based on the user features and the process capabilities of the plurality of training devices. However, Bhatt teaches similar feature (see e.g., Para [17][25][27]-[29] – “model training manager 140 can adjust a performance metric of client device 120 to reflect degraded or lesser expected performance from client device 120 and assign smaller training data sets to client device 120 for processing.”” In some embodiments, parameter service 144 can adjust the performance level of a client device 120 in response to a failure to receive a response from the client device 120 within the timeout period. By adjusting the performance level of the client device downwards to reflect degraded performance at client device 120, parameter service 144 can adjust the amount of data that may be assigned to the client device 120 for subsequent processing such that the client device 120 can continue to participate in training a machine learning model associated with a specific job. In some embodiments, if parameter service 144 adjusts the performance level of a client device 120 below a minimum performance level defined for a job, parameter service 144 can remove the client device 120 from the set of devices participating in training the machine learning model associated with the job.”) Both Nakayama and Bhatt are directed to machine learning methods. Accordingly, it would have been obvious to the skilled artisan before the effective filing date of the claimed invention having Nakayama and Bhatt in front of them to modify the system of Nakayama to include the above feature. The motivation to combine Nakayama and Bhatt comes from Bhatt. Bhatt discloses the motivation to group devices based on performance so that training speed and reliability can be improved (see e.g., Para [17][25][27]-[29]). The motivation to combine also applies to the following dependent claims.
With respect to dependent claim 2, the modified Nakayama teaches acquiring user device information from a plurality of user devices; and selecting the plurality of training devices from the plurality of user devices based on the user device information (see e.g., Para [32]-[49][97][144][145] – “The term “agent”, “device”, or “client” as used herein means a system with distributed learning environment such as local edge server, device, tablet, among others, in order to train machine learning models locally and send them to an associated aggregator.” “multiple aggregators coupled to the communication network and each uniquely associated with the agents, each aggregator comprising [0038] a model collector collecting the local machine learning models from the associated agents; [0039] a memory storing the collected local machine learning models; and [0040] a processor creating a cluster machine learning model from the collected local machine learning models,” Nakayama does not expressly show acquiring information from user devices. However, it would have been obvious to include this feature, because Nakayama requires the determination of “association” between certain devices to form group).
With respect to dependent claim 3, the modified Nakayama teaches the transmitting the model to be trained to the training devices comprises: transmitting first information corresponding to the feature extraction layers for extracting user features to the plurality of training devices; determining pre-trained groups of the plurality of training devices, respectively, based on a pre-trained grouping result; and transmitting, to the plurality of training devices, second information corresponding to the prediction layers based on the pre-trained groups (see e.g., Para [98][125] – “The group of cluster aggregators communicate with other group(s) of aggregators periodically to exchange their semi-global machine learning models to create a global machine learning model. This communication enables each user to utilize the training results of the users in other groups by receiving a most-updated AI model that approximates a consistent global AI model.”).
With respect to dependent claim 5, the modified Nakayama teaches the classifying the plurality of training devices into the at least one group further comprises: clustering the plurality of training devices based on the user features of the respective training devices to obtain at least one first level group; and classifying, for each of the at least one first level group, the respective training devices based on the process capabilities of the respective training devices within the at least one first level group to obtain at least one second level group, the obtained respective second levels of groups serving as a grouping result (see e.g., Para [17][25][27]-[29] – “model training manager 140 can adjust a performance metric of client device 120 to reflect degraded or lesser expected performance from client device 120 and assign smaller training data sets to client device 120 for processing.”” In some embodiments, parameter service 144 can adjust the performance level of a client device 120 in response to a failure to receive a response from the client device 120 within the timeout period. By adjusting the performance level of the client device downwards to reflect degraded performance at client device 120, parameter service 144 can adjust the amount of data that may be assigned to the client device 120 for subsequent processing such that the client device 120 can continue to participate in training a machine learning model associated with a specific job. In some embodiments, if parameter service 144 adjusts the performance level of a client device 120 below a minimum performance level defined for a job, parameter service 144 can remove the client device 120 from the set of devices participating in training the machine learning model associated with the job.” Capability based classification corresponds to the second level grouping).
With respect to dependent claim 6, the modified Nakayama teaches the performing global federated aggregation based on the first parameters of the respective training devices comprises: weighted averaging the first parameters of the respective training devices to obtain the global federated aggregation result (see e.g., Para [137]-[142]).
With respect to dependent claim 7, the modified Nakayama teaches the performing intra-group federated aggregation on the second parameters of the respective training devices in the group in each of the at least one group comprises: weighted averaging the second parameters of the respective training devices in the group in each of the at least one group to obtain the intra-group federated aggregation result (see e.g., Para [137]-[142]).
With respect to dependent claim 8, the modified Nakayama teaches the training method further comprises: updating the grouping result (see e.g., Para [97]-[98]).
With respect to dependent claim 9, the modified Nakayama teaches the updating the grouping result comprises: calculating a similarity between each of the plurality of training devices and each of the at least one group, respectively; and updating the grouping result based on the similarity (see e.g., Para [135]-[138).
With respect to dependent claim 10, the modified Nakayama teaches the prediction model is configured to predict user attribute information (see e.g., Para [122]-[125] and [145]).
With respect to dependent claim 11, the modified Nakayama teaches repeatedly performing the operations of: receiving the model parameter, performing the global federated aggregation and the intra-group federated aggregation, and transmitting the global federated aggregation result and the intra-group federated aggregation result until end of training (see e.g., Para [144] – “hen a personalized model is obtained by combining the local model and the global model using the personalization rate, where the personalization rate measures the extent to which the personalized model mixes the local and the global models. Then the personalized model is tested to check whether a certain performance criteria is met. If the criteria is not met, the global model is updated using and a new round of training is started. This procedure repeats until the performance criterion is satisfied, in other words, the personalized model generalizes sufficiently well for the local dataset distribution. Finally, the personalized model for each agent is output.”).
With respect to independent claim 12, the modified Nakayama teaches a prediction model training method, which is performed by a server (see e.g., Para [7][97] – “The approximated global model created through this global model synthesis process is called a semi-global model. The term “cluster aggregator” or “CA” or “server” as used herein means a system that aggregates, via a communication network, artificial intelligence (AI) models that are trained at multiple agents (defined below) and creates a cluster machine learning model from the aggregated AI models. The aggregator serves as a federated learning (FL) server. The term “agent”, “device”, or “client” as used herein means a system with distributed learning environment such as local edge server, device, tablet, among others, in order to train machine learning models locally and send them to an associated aggregator.”), the method comprising:
transmitting a model to be trained to a plurality of training devices, the model to be trained comprising feature extraction layers configured to extract user features and prediction layers configured to perform information prediction (see e.g., Para [12]-[16]- “Models trained in advance with prepared training data Comparisons only against the static base model Static models are trained for limited set of real world scenarios. Single model is deployed”);
classifying the plurality of training devices into at least one group based on the user features extracted by the plurality of training devices and transmitting a grouping result to the plurality of training devices; receiving model parameters obtained by the plurality of training devices training the model to be trained, wherein the model parameters comprise first parameters corresponding to the feature extraction layers (see e.g., Para [32]-[40][145]- “multiple aggregators coupled to the communication network and each uniquely associated with the agents, each aggregator comprising a model collector collecting the local machine learning models from the associated agents”” The personalization can be extendedly interpreted as a model aggregation for each group of users who share a similar behavioral pattern. The group-level model management and preparation virtually cluster all the users into multiple groups by incorporating a feature vector-based clustering method. This enables the customization and advanced control of ML models distributed by aggregators for different types of users.”);
performing global federated aggregation on the first parameters of the respective training devices to obtain a global federated aggregation result (see e.g., Fig. 10 Para [50][110][137]-[144]- “the system may further comprise a model repository storing the global machine learning models previously created by the system and meta-data indicating tasks used for training the respective global machine learning models”);
transmitting the global federated aggregation result to the plurality of training devices so that the plurality of training devices update the feature extraction layers based on the global federated aggregation result (see e.g., Para [141]-[144]- “he personalization can be extendedly interpreted as a model aggregation for each group of users who share a similar behavioral pattern. The group-level model management and preparation virtually cluster all the users into multiple groups by incorporating a feature vector-based clustering method. This enables the customization and advanced control of ML models distributed by aggregators for different types of users.”); wherein the classifying the plurality of training devices into the at least one group comprises: acquiring process capabilities of each of the plurality of training devices; and classifying the plurality of training devices into one or more groups among the at least one group based on the user features and the process capabilities of the plurality of training devices (see e.g. Bhatt Para [17][25][27]-[29] and discussion above with respect to claim 1.)
With respect to dependent claim 13, the modified Nakayama teaches a prediction model training method, which is performed by a server (see e.g., Para [7][97] – “The approximated global model created through this global model synthesis process is called a semi-global model. The term “cluster aggregator” or “CA” or “server” as used herein means a system that aggregates, via a communication network, artificial intelligence (AI) models that are trained at multiple agents (defined below) and creates a cluster machine learning model from the aggregated AI models. The aggregator serves as a federated learning (FL) server. The term “agent”, “device”, or “client” as used herein means a system with distributed learning environment such as local edge server, device, tablet, among others, in order to train machine learning models locally and send them to an associated aggregator.”), comprising: receiving model parameters obtained by a plurality of training devices in a first group, among at least one group, training the model to be trained, the model to be trained comprising feature extraction layers configured to extract user features and prediction layers configured to perform information prediction (see e.g., Para [32]-[40][145]- “multiple aggregators coupled to the communication network and each uniquely associated with the agents, each aggregator comprising a model collector collecting the local machine learning models from the associated agents”” The personalization can be extendedly interpreted as a model aggregation for each group of users who share a similar behavioral pattern. The group-level model management and preparation virtually cluster all the users into multiple groups by incorporating a feature vector-based clustering method. This enables the customization and advanced control of ML models distributed by aggregators for different types of users.”), and the model parameters comprise second parameters corresponding to the prediction layers (see e.g., Para [12]-[16]- “Models trained in advance with prepared training data Comparisons only against the static base model Static models are trained for limited set of real world scenarios. Single model is deployed”); performing intra-group federated aggregation on the second parameters of the plurality of training devices in the first group to obtain an intra-group federated aggregation result (see e.g., Para [32]-[40][145] [144]- “this two model approach, for each agent we first randomly initialize the two models”); and transmitting the intra-group federated aggregation result to the plurality of training devices in the first group so that the plurality of training devices update the prediction layers based on the intra-group federated aggregation result (see e.g., Para [141]-[144]- “he personalization can be extendedly interpreted as a model aggregation for each group of users who share a similar behavioral pattern. The group-level model management and preparation virtually cluster all the users into multiple groups by incorporating a feature vector-based clustering method. This enables the customization and advanced control of ML models distributed by aggregators for different types of users.”), wherein the first group is determined based on user features and process capabilities of the plurality of the plurality of training devices (see e.g. Bhatt Para [17][25][27]-[29] and discussion above with respect to claim 1.)
With respect to independent claim 14, the modified Nakayama teaches a prediction model training method, which is performed by a training device, the method comprising: receiving a model to be trained from a server (see e.g., Abstract Para [7][97]- – “The approximated global model created through this global model synthesis process is called a semi-global model. The term “cluster aggregator” or “CA” or “server” as used herein means a system that aggregates, via a communication network, artificial intelligence (AI) models that are trained at multiple agents (defined below) and creates a cluster machine learning model from the aggregated AI models. The aggregator serves as a federated learning (FL) server. The term “agent”, “device”, or “client” as used herein means a system with distributed learning environment such as local edge server, device, tablet, among others, in order to train machine learning models locally and send them to an associated aggregator.”), the model to be trained comprising feature extraction layers configured to extract user features and prediction layers configured to perform information prediction (see e.g., Para [12]-[16] [21]-[29][122][123]- “Models trained in advance with prepared training data Comparisons only against the static base model Static models are trained for limited set of real world scenarios. Single model is deployed”);
extracting a user feature using the feature extraction layers in the model to be trained, and transmitting the extracted user feature to the server to classify the training device into one of at least one group based on the user features (see e.g., Para [32]-[40][145]- “multiple aggregators coupled to the communication network and each uniquely associated with the agents, each aggregator comprising a model collector collecting the local machine learning models from the associated agents”” The personalization can be extendedly interpreted as a model aggregation for each group of users who share a similar behavioral pattern. The group-level model management and preparation virtually cluster all the users into multiple groups by incorporating a feature vector-based clustering method. This enables the customization and advanced control of ML models distributed by aggregators for different types of users.”);
training the model to be trained, and transmitting model parameters obtained by training to the server (see e.g., Abstract Para [7][21][97]- – “The approximated global model created through this global model synthesis process is called a semi-global model. The term “cluster aggregator” or “CA” or “server” as used herein means a system that aggregates, via a communication network, artificial intelligence (AI) models that are trained at multiple agents (defined below) and creates a cluster machine learning model from the aggregated AI models. The aggregator serves as a federated learning (FL) server. The term “agent”, “device”, or “client” as used herein means a system with distributed learning environment such as local edge server, device, tablet, among others, in order to train machine learning models locally and send them to an associated aggregator.”), wherein the model parameters comprise first parameters corresponding to the feature extraction layers and second parameters corresponding to the prediction layers (see e.g., Para [21]-[29][122][123]- “The network comprises three sequentially connected blocks of layers. (a) The embedding block 80 takes the process model state as an input and converts it into a common representation by accounting for the heterogeneity of process models. The embedding block 80 additionally predicts process variables that are not part of the process state but can be descriptive of the process performance and other metrics that an operator cares about. Such variables are determined by each agent's operator. (b) The inference block 90, whose parameters are aggregated through the federated learning process, uses the common representation of the input to produce an output. Since the inference block 90 is agnostic to process model variations, it can be generated by aggregating inference blocks 90 across the network. (c) The transfer block 100 converts the common representation of the output into an output value understood by the particular process model. The transfer block 100 can also predict process variables aside from ones assigned to the embedding block 80. Process variables are used by inference and transform blocks 80, 90 to calculate the output.”);
receiving a global federated aggregation result and an intra-group federated aggregation result from the server (see e.g., Para [141]-[144]- “the personalization can be extendedly interpreted as a model aggregation for each group of users who share a similar behavioral pattern. The group-level model management and preparation virtually cluster all the users into multiple groups by incorporating a feature vector-based clustering method. This enables the customization and advanced control of ML models distributed by aggregators for different types of users.”),
wherein the global federated aggregation result is obtained by the server performing global federated aggregation on the first parameters of a plurality of training devices including the training device, and the intra-group federated aggregation result is obtained by the server performing intra-group federated aggregation on the second parameters of one or more the plurality of training devices in the one of the at least one group (see e.g., Para [21]-[29][122][123]- “The network comprises three sequentially connected blocks of layers. (a) The embedding block 80 takes the process model state as an input and converts it into a common representation by accounting for the heterogeneity of process models. The embedding block 80 additionally predicts process variables that are not part of the process state but can be descriptive of the process performance and other metrics that an operator cares about. Such variables are determined by each agent's operator. (b) The inference block 90, whose parameters are aggregated through the federated learning process, uses the common representation of the input to produce an output. Since the inference block 90 is agnostic to process model variations, it can be generated by aggregating inference blocks 90 across the network. (c) The transfer block 100 converts the common representation of the output into an output value understood by the particular process model. The transfer block 100 can also predict process variables aside from ones assigned to the embedding block 80. Process variables are used by inference and transform blocks 80, 90 to calculate the output.”); and
updating the feature extraction layers based on the global federated aggregation result, and updating the prediction layers based on the intra-group federated aggregation result (see e.g., Para [141]-[144]- “the personalization can be extendedly interpreted as a model aggregation for each group of users who share a similar behavioral pattern. The group-level model management and preparation virtually cluster all the users into multiple groups by incorporating a feature vector-based clustering method. This enables the customization and advanced control of ML models distributed by aggregators for different types of users.”) wherein the at least one group is determined based on user features extracted by the plurality of training devices and process capabilities of the plurality of training devices (see e.g. Bhatt Para [17][25][27]-[29] and discussion above with respect to claim 1.).
With respect to dependent claim 16, the modified Nakayama teaches the performing global federated aggregation based on the first parameters comprises performing global federated aggregation based on the first parameters of the plurality of training devices across all of the at least one group to obtain the global federated aggregation result (see e.g., Para [41]-[43][97]-[99] and Bhatt Para [20] – “Multiple cluster aggregators are coupled via the communication network to form a group of cluster aggregators and exchange their cluster machine learning models with each other to create a semi-global machine learning model. The group of cluster aggregators communicate with other group(s) of aggregators periodically to exchange their semi-global machine learning models to create a global machine learning model.”).
With respect to dependent claim 17, the modified Nakayama teaches the performing intra-group federated aggregation for each of the at least one group comprises performing intra- group federated aggregation for each of the at least one group based on the second parameters of one or more of the plurality of training devices only within the respective group to obtain the intra-group federated aggregation result (see e.g., Para [101]-[116] and Bhatt Para [20][26] – “Each agent 20 belongs to a cluster that is managed by a cluster aggregator (CA). The client uploads a machine learning (ML) model trained by its own local data to the corresponding CA 30 where a model aggregation algorithm is performed to create a cluster model. “ “A cluster model is formed by aggregating all the accepted models. The CA pushes the cluster model into a database and simultaneously accesses some databases to retrieve cluster models formed by other CAs. A semi-global model, which will be used in the agents belonging to the CA, is synthesized from the cluster models.”).
Claim 15 is rejected under 35 U.S.C. 103 as being unpatentable over Nakayama in view of Bhatt and further in view of Choe et al. (hereinafter Choe) U.S. Patent Publication No. 2021/0250510.
With respect to independent claim 15, Nakayama teaches an information prediction method, which is executed by a user device, the method comprising: receiving parameters of feature extraction layers of a prediction model and first central point information corresponding to a first group (see e.g., Para [141]-[144]- “the personalization can be extendedly interpreted as a model aggregation for each group of users who share a similar behavioral pattern. The group-level model management and preparation virtually cluster all the users into multiple groups by incorporating a feature vector-based clustering method. This enables the customization and advanced control of ML models distributed by aggregators for different types of users.”), among at least one group, the feature extraction layers configured to extract user features, and the first information representing an average user feature of user devices within the first group (see e.g., Para [137] [138] [142] - “The standard federated learning typically assumes that all user's data come from a similar distribution so that every single agent can benefit from other's data by participating in the federated learning process. However, if the distribution of an agent's dataset drifts far away from the average distribution among all the other agents, the global model trained from federated learning might be ineffective to this agent. To resolve this problem, it is necessary to find a way to better utilize the generalization ability of the global model while not compromising the model performance for the local distribution. This motivates an introduction of the personalization module in the system of the present disclosure.”)
obtaining the prediction model corresponding to the user device based on the feature extraction layers, the first central point information and user data of the user device (see e.g., Fig. 4 and Para [110]-[125]); and predicting information using the obtained prediction model (see e.g., Para [122]-[123]) wherein the obtaining the prediction model comprises acquiring a process capability of the user device and selecting a group from the at least one group based on the user features and the process capability of the user device (see e.g. Bhatt Para [17][25][27]-[29] and discussion above with respect to claim 1.).
Nakayama does not expressly show the first center point information. However, Nakayama expressly indicates that average value is used to determine drift. Furthermore, Choe teaches similar feature (see e.g., Para [80]-[84] – “The electronic device 101 obtains a center point of the extracted features and the weight moving average is used to smooth the coordinates along the temporal image sequence 616”). Both Nakayama and Choe are directed to machine learning methods. Accordingly, it would have been obvious to the skilled artisan before the effective filing date of the claimed invention having Nakayama and Choe in front of them to further modify the modified system of Nakayama to include the above feature. The motivation to combine Nakayama and Choe comes from Choe. Choe discloses the motivation to determine a center point for feature extraction so that boundary can be defined for features (see e.g., Para [80]-[82]).
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 PEIYONG WENG whose telephone number is (571)270-1660. The examiner can normally be reached on Mon.-Fri. 8 am to 5 pm.
If attempts to reach the examiner by telephone are unsuccessful, the examiner's supervisor, Matthew Ell, can be reached on (571) 270-3264. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://portal.uspto.gov/external/portal. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free).
/PEI YONG WENG/ Primary Examiner, Art Unit 2141
Prosecution Insights