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 .
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
Claim(s) 1-4 is/are rejected under 35 U.S.C. 102 (a)(1) as being anticipated by Yang et al. (U.S. pat. 12,182,673 B1)
With respect to claim 1, Yang et al. discloses a model configuration method implemented by an apparatus, comprising:
receiving a first model request message from a first access network device (i.e., “ a request for the compatible machine learning model may be sent to model generation system 110, as discussed below with regard to FIGS. 3 and 5, which may then perform a network architecture search using weights determined for the compatible machine learning model and a reward function that enforces compatibility between machine learning model 132 and a perspective compatible machine learning model to find, train, and provide compatible machine learning model 122.”(col. 3, lines 35-44) and “Object recognition service 210 may implement interface 211 to allow clients (e.g., client(s) 250 or clients implemented internally within provider network 200, such as a client application hosted on another provider network service like an event driven code execution service or virtual compute service) to send search requests, add objects, such as image data (which may be found in various types of media, such as still images or video data) or other data (e.g., text/character strings for natural language processing or audio data), requests to create a query model within resource constraints, or requests to search generated object index(es) 232 using a query model that includes an embedding as a reference for performing a search”(col. 5, lines 10-22) and query model or text or character string, etc. are message as claimed invention);
determining a first model based on the first model request message and a first mapping relationship (i.e., “ request for the compatible machine learning model may be sent to model generation system 110, as discussed below with regard to FIGS. 3 and 5, which may then perform a network architecture search using weights determined for the compatible machine learning model and a reward function that enforces compatibility between machine learning model 132 and a perspective compatible machine learning model to find, train, and provide compatible machine learning model 122.”(col. 3, lines 33-42) and “which may utilize machine learning model 132 to map objects 142 (e.g., images) into vector space 140. In order to support other systems, such as system 120, being able to access the objects 142 efficiently, a separate model may be utilized.”(col. 3, lines 19-25)), wherein the first mapping relationship comprises a mapping relationship between a model (i.e., “which may utilize machine learning model 132 to map objects 142 (e.g., images) into vector space 140. In order to support other systems, such as system 120, being able to access the objects 142 efficiently, a separate model may be utilized.”(col. 3, lines 19-25) and “the resource constraint may be selected from a provided set of constraints, devices, or scenarios that map to a resource constraint (e.g., selecting edge device type X maps to resource constraint Y). In some embodiments, a gallery model may be specified (e.g., out of a multiple different models used for different galleries). In some embodiments, a default model may be automatically determined for a request (e.g., a visual search service may automatically use the machine learning model used for indexing objects at the visual search service).”(col. 7, lines 46-58) and col. 9, lines 1-9), a model application scenario (i.e., “ In such scenarios, model generation for edge devices 216 may send, store, or otherwise provide the generated query model 326 to other service deployment resource….As indicated at 324, a generated query model could also be stored in storage service(s) 230, which may be later accessed to retrieve or obtain the generated query model. As indicated at 322, in some scenarios the generated query model may be returned via interface 211.”(col. 5, lines 9-15) and “the resource constraint may be selected from a provided set of constraints, devices, or scenarios that map to a resource constraint (e.g., selecting edge device type X maps to resource constraint Y). In some embodiments, a gallery model may be specified (e.g., out of a multiple different models used for different galleries)”(col. 7, lines 48-54)), and a model function (i.e., “a request for the compatible machine learning model may be sent to model generation system 110, as discussed below with regard to FIGS. 3 and 5, which may then perform a network architecture search using weights determined for the compatible machine learning model and a reward function that enforces compatibility between machine learning model 132 and a perspective compatible machine learning model to find, train, and provide compatible machine learning model 122.”(col. 3, lines 35-43 )), or the first mapping relationship comprises a mapping relationship between a model (i.e., “which may utilize machine learning model 132 to map objects 142 (e.g., images) into vector space 140. In order to support other systems, such as system 120, being able to access the objects 142 efficiently, a separate model may be utilized.”(col. 3, lines 19-25) and “the resource constraint may be selected from a provided set of constraints, devices, or scenarios that map to a resource constraint (e.g., selecting edge device type X maps to resource constraint Y). In some embodiments, a gallery model may be specified (e.g., out of a multiple different models used for different galleries). In some embodiments, a default model may be automatically determined for a request (e.g., a visual search service may automatically use the machine learning model used for indexing objects at the visual search service).”(col. 7, lines 46-58) and col. 9, lines 1-9), a model performance level (i.e., “a new version of a deep neural network may be updated while an older version is still in use, or in another example machine learning models with different computational complexities utilizing different amounts of resources are deployed to different systems or devices (e.g., a large, resource intensive machine learning model deployed at cloud services or other large-scale systems and a different smaller machine learning model deployed on edge devices to perform the same task).”(col. 2, lines 10-20) or “ services which can utilize compatible machine learning models to perform various tasks.”(col. 2, lines 31-32) or “ As indicated at 630, a number of possible architectures identified as part of the search may be selected according to respective performance as indicated by the reward function. For example, a top five performing architectures may be selected, in some embodiments”(col. 13, lines 25-30)), and a model function (i.e., “Generating machine learning models within resource constraints rely upon compatibility awareness allows for a better trade off accuracy and complexity, by optimizing over both model parameters and architecture. In various embodiments, weight-based compatibility may be combined with various reward functions to achieve compatibility-aware architecture search”(col. 8, lines 40-45)); and
sending information about the first model to the first access network device (i.e., “ a client 250 may utilize a query model 252 (e.g., which may be generated by model generation for edge devices 216) in order to send search requests for various objects in an object index 232, which may be sent via interface 211 and handed by index search 212.”(col. 6, line 59-63) and step 530 of fig. 5).
With respect to claim 2, Yang et al. discloses wherein the first model request message indicates a model application scenario (i.e., “A query model creation request 310 may be received via interface 211 at model generation for edge devices 216. The request 310 may specify various features for generating the model including a resource constraint, gallery model with which the generated model is to be compatible and/or a destination for the generated model. … the resource constraint may be selected from a provided set of constraints, devices, or scenarios that map to a resource constraint (e.g., selecting edge device type X maps to resource constraint Y).”(col. 7, lines 37-52)) and a model function of the first model (i.e., “ a request for the compatible machine learning model may be sent to model generation system 110, as discussed below with regard to FIGS. 3 and 5, which may then perform a network architecture search using weights determined for the compatible machine learning model and a reward function that enforces compatibility between machine learning model 132 and a perspective compatible machine learning model to find, train, and provide compatible machine learning model 122.”(col. 3, lines 35-44)).
With respect to claim 3, Yang et al. discloses wherein the first model request message indicates a model function scenario (i.e., “A query model creation request 310 may be received via interface 211 at model generation for edge devices 216. The request 310 may specify various features for generating the model including a resource constraint, gallery model with which the generated model is to be compatible and/or a destination for the generated model. … the resource constraint may be selected from a provided set of constraints, devices, or scenarios that map to a resource constraint (e.g., selecting edge device type X maps to resource constraint Y).”(col. 7, lines 37-52)) and a model performance level of the first model (i.e., “By including the architecture in the design of metrically compatible model, a search among architectures Qq for the query images to maximize efficiency while ensuring that performance remains close to that of using a same larger model for both the query and gallery.”(col. 3, lines 4-10)), or indicates a model function and a model performance indicator of the first model.
With respect to claim 4, Yang et al. discloses wherein the information about the first model indicates at least one of the following of the first model: a model index (i.e., “This specification begins with a general description of a provider network that implements multiple different services, including a machine learning model-based search service, which may generate an index for various types of objects (e.g., video, images, audio, etc.) using a machine learning model and supports requests to search or otherwise access the objects using a query machine learning model.”(col. 3, lines 50-57)), model structure information, a model parameter, a model input format, a model output format (i.e., “ one data storage service 230 may be an object-based data store that allows for different data objects of different formats or types of data, such as structured data (e.g., database data stored in different database schemas), unstructured data (e.g., different types of documents or media content), or semi-structured data (e.g., different log files, human-readable data in different formats like JavaScript Object Notation (JSON) or Extensible Markup Language (XML)) to be stored and managed according to a key value or other unique identifier that identifies the object.”(col. 6, lines 25-35)), a model performance indicator, a model application scenario, a model function, or a training parameter.
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 5-7 are rejected under 35 U.S.C 103(a) as being unpatentable over yang et al.( U.S. pat. 12,182,673 B1) in view of Ha. (U.S. Pub. 2023/0214773 A1)
With respect to claim 5, Yang et al. discloses all limitations recited in claim 1 except for receiving first feedback information from the first access network device and updating the first model based on the first feedback information. However, Ha discloses receiving first feedback information from the first access network device and updating the first model based on the first feedback information (i.e., “Accordingly, the predictive model may be dynamically updated based on feedback acquired by monitoring inventory units at one or more sites. In particular, the predictive model can be dynamically updated with counts of inventory units to continuously adjust the reorder point and/or reorder quantity values, thereby providing a self-optimizing inventory management system.”(0058)). It would have been obvious for a person of ordinary skill in the art, before the effective filing date of the claimed invention, to include Ha’s feature in order to have easy and increase accurate of the predict model for the stated purpose has been well known in the art as evidenced by teaching of Ha (0002, 0004).
With respect to claim 6, Ha discloses the method according to claim 5, wherein the first feedback information indicates actual accuracy of prediction information output by the first model (i.e., “ the machine learning model may be updated based on actual results (e.g., comparing the predicted outputs to the actual service level for one or more sites and/or the actual total annual cost for inventory for a region) to continuously retrain the machine learning model, thus improving the accuracy of the predictive modeling performed in accordance with present invention embodiments over time.”(0025)), and the updating the first model based on the feedback information comprises: when the actual accuracy of the prediction information output by the first model is less than a threshold (i.e., “ In response to determining that the service level does not satisfy a threshold value, the reorder point is adjusted. The adjusted reorder point is determined to yield a service level, according to the predictive model, that satisfies” (abstract) the threshold value.’(abstract)), retraining the first model, and updating the first model (i.e., “”the machine learning model may be updated based on actual results (e.g., comparing the predicted outputs to the actual service level for one or more sites and/or the actual total annual cost for inventory for a region) to continuously retrain the machine learning model, thus improving the accuracy of the predictive modeling performed in accordance with present invention embodiments over time.”(0025)).
With respect to claim 7, Ha discloses the method the method according to claim 5, wherein the first feedback information indicates a scenario change indication (i.e., “ the predictive model may be dynamically updated based on feedback acquired by monitoring inventory units at one or more sites. In particular, the predictive model can be dynamically updated with counts of inventory units to continuously adjust the reorder point and/or reorder quantity values, thereby providing a self-optimizing inventory management system.”(0058)), a model performance level change indication, or a model performance indicator change indication (i.e., “Data relating to performing predictive modeling (e.g., data relating to counts, transfer times, lead times, prices, and/or consumption of inventory units, predictive model data, etc.) may be stored within any conventional or other data structures (e.g., files, arrays, lists, stacks, queues, records, etc.) and may be stored in any desired storage unit (e.g., database, data or other repositories, queue, etc.). ”(0069)), and the updating the first model based on the first feedback information (i.e., “ the predictive model may be dynamically updated based on feedback acquired by monitoring inventory units at one or more sites. In particular, the predictive model can be dynamically updated with counts of inventory units to continuously adjust the reorder point and/or reorder quantity values, thereby providing a self-optimizing inventory management system.”(0058)), comprises: selecting a second model for a terminal device based on the scenario change indication((i.e., “ the predictive model may be dynamically updated based on feedback acquired by monitoring inventory units at one or more sites. In particular, the predictive model can be dynamically updated with counts of inventory units to continuously adjust the reorder point and/or reorder quantity values, thereby providing a self-optimizing inventory management system.”(0058)), the model performance level change indication, or the model performance indicator change indication ((i.e., “ the predictive model may be dynamically updated based on feedback acquired by monitoring inventory units at one or more sites. In particular, the predictive model can be dynamically updated with counts of inventory units to continuously adjust the reorder point and/or reorder quantity values, thereby providing a self-optimizing inventory management system.”(0058)); and father, Yang et al. discloses sending information about the second model to the first access network device, wherein the second model is for updating the first model (i.e., “ machine learning model-based search service 210 may implement index generation/update 214. Index generation/update 214 may add objects to or create new object catalogs using a corresponding machine learning model (e.g., a gallery model as discussed above), which may generate representations of objects as embeddings. … Requests may be received via interface 211 to create new object indexes, including requests that specify a particular machine learning model and/or set of objects to include in the index. Index generation/update 214 may receive these requests, as well as requests to add objects to an index, and apply the corresponding machine learning model to generate and store the embeddings in object index(es) 234. Index generation/update 214 may utilize training data 236, in some embodiments, to train the machine learning model(s) 234 used by index generation/update 214”(col. 5, lines 53-67 and col. 6, lines 1-7)) (Same motivation as claim 4 above).
With respect to claims 8-20, the claims 8-20 are rejected as same of claims 1-7 above because the set of claims 8-20 are similar with set of claims 1-7 but different form.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to HUNG T VY whose telephone number is (571)272-1954. The examiner can normally be reached M-F 8-5.
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, Tony Mahmoudi can be reached at (571)272-4078. 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.
/HUNG T VY/Primary Examiner, Art Unit 2163 September 28, 2026
