NON-FINAL REJECTION, FIRST DETAILED ACTION
Status of Prosecution
The present application, 18/394,697 filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
The application was filed in the Office on Dec. 22, 2023.
Claims 1-20 are pending. Claims 1-5, 8-16 and 19-20 are rejection. Claims 6-7 and 17-18 are objected to. Claims 1, 12 and 20 are independent.
Status of Claims
Claims 6-7, 17-18, are objected to.
Claims 1-3, 8-14 and 19-20 are rejected under 35 U.S.C. §103 as being unpatentable over non-patent literature, Li et al. (“Li”), “Effective Document Labeling with Very Few Seed Words: A Topic Modeling Approach,” published in 2016 in view of .non-patent literature, Meng et al. (“Meng”), “Generating Training Data with Language Models: Towards Zero-Shot Language Understanding,” published in 2022 in further view of non-patent literature, Naeem et al. (“Naeem”), “I2MVFormer: Large Language Model Generated Multi-View Document Supervision for Zero-Shot Image Classification,” published in 2023.
Claims 4-5 and 15-16 are rejected under 35 U.S.C. §103 as being unpatentable over Li in view of Meng in view of Naeelm and in further view of non-patent literature, Liu et al. (“Liu”), “Improving Embedding-based Large-scale Retrieval via Label Enhancement,” published in in 2021.
Objection and Allowable Subject Matter
Claims 6-7, 17, 18, are objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening 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.
A.
Claims 1-3, 8-14 and 19-20 are rejected under 35 U.S.C. §103 as being unpatentable over non-patent literature, Li et al. (“Li”), “Effective Document Labeling with Very Few Seed Words: A Topic Modeling Approach,” published in 2016 in view of .non-patent literature, Meng et al. (“Meng”), “Generating Training Data with Language Models: Towards Zero-Shot Language Understanding,” published in 2022 in further view of non-patent literature, Naeem et al. (“Naeem”), “I2MVFormer: Large Language Model Generated Multi-View Document Supervision for Zero-Shot Image Classification,” published in 2023.
As to Claim 1, Li teaches: A computer-implemented method comprising:
obtaining a first set of seed data objects based on a first identified desired attribute (Li: Fig. 2, sec. 4.2, seed words selection may be conducted in a manner that they are of high quality (i.e. an identified desired attribute);
retrieving a first plurality of candidates from a database of data objects based on similarity to the first set of seed data objects (Li: Fig. 2, the unlabeled documents (i.e. first plurality of candidates) are assigned predicted labels based on topic inference and other scoring (i.e. similarity);
annotating the first plurality of candidates based on a list of defined labels (Li: Fig. 2, Abstract, the documents are labeled).
PNG
media_image1.png
392
497
media_image1.png
Greyscale
Li may not explicitly teach: applying an embedding transformation to each of the seed data objects in the first set of seed data objects to create a first modified set of seed data objects;
retrieving a first plurality of candidates from a database of data objects based on similarity to the first modified set of seed data objects;
annotating the first plurality of candidates based on a list of defined labels to create a training dataset including the first plurality of annotated candidates; and
training a machine learning model using the training dataset.
Meng teaches in general concepts related to generating training data with language models using zero-shot language understanding (Meng: Abstract). Specifically, pretrained language models (PLMs) are used (SuperGen), wherein the training data are created bia a unidirectional PLM (a generator) which generates class-conditioned texts guided by label-descriptive prompts. (Meng: Sec. 1). Then a bidirectional PLM (a classifier) is fine-tuned on the generated texts to perform task(Meng: Sec. 1).The generator may be one that uses contextualized embeddings given by a transformer encoder (Meng: Sec. 3.1, the embedding vector h). Therefore, each of the generated texts are defined within an embedding space and encoded thus.
It would have been obvious to a person having ordinary skill in the art at a time before the effective filing date of the application to have modified the Li disclosures and teachings by applying an embedding transformation to the seed words by use of the PLMs as taught and suggested by Meng. Such a person would have been motivated to do so with a reasonable expectation of success to allow for an efficient and optimized approach to better generating the training data.
Li and Meng may not explicitly teach: using a large language model (LLM), annotating the first plurality of candidates based on a list of defined labels to create a training dataset including the first plurality of annotated candidates,
Naeem teaches in general concepts related to using large language models (LLMs) to provide text supervision for a zero-shot image classification model (Naeem: Abstract). Specifically, the LLM is conditioned with a few text description from different annotators (i.e. the seed data objects)(Naeem: Abstract). The LLMs are used to generate multiple views of a class, exploiting the storage of knowledge from multiple sources and to generate multi-view text descriptions of object classes using a pretrained LLM (Naeem: Sec. 3).
It would have been obvious to a person having ordinary skill in the art at a time before the effective filing date of the application to have modified the Li-Meng disclosures and teachings by utilizing an LLM as one of the PLM’s of Meng as taught by Naeem. Such a person would have been motivated to do so with a reasonable expectation of success to allow for the exploiting of the storage of knowledge from multiple sources and to generate multi-view text descriptions of object classes using a pretrained LLM (Naeem: Sec. 3).
As to Claim 2, Li, Meng and Naeem teach the elements of claim 1.
Meng further teaches: wherein the database of data objects comprises a plurality of embeddings defining an embedding space and the first plurality of candidates represents a subset of the embeddings (Meng: Sec. 3.1, the embedding vector h).
As to Claim 3, Li, Meng and Naeem teach the elements of claim 2.
Meng further teaches: wherein applying the embedding transformation to each of the seed data objects in the first set of seed data objects comprises: encoding each of the seed data objects as a respective seed embedding within the embedding space (Meng: Sec. 3.1, the embedding vector h).
As to Claim 8, Li, Meng and Naeem teach the elements of claim 1.
Meng further teaches: wherein obtaining the first set of seed data objects comprises: querying an external database to obtain the first set of seed data objects (Meng: Sec. 3.1, the corpus D (e.g. Wikipedia) may be reasonably understood to be an external database).
As to Claim 9, Li, Meng and Naeem teach the elements of claim 1.
Naeem further teaches: wherein the first set of seed data objects is a set of images (Naeem: Abstract, the images that are considered).
As to Claim 10, Li, Meng and Naeem teach the elements of claim 1.
Naeem further teaches: wherein the first set of seed data objects is metadata of a set of images (Naeem: Abstract text descriptions, which may be metadata).
As to Claim 11, Li, Meng and Naeem teach the elements of claim 1.
Li further teaches: wherein the first set of seed data objects is a set of textual descriptions (Li: Fig. 2, sec. 4.2, seed words selection which are text).
As to Claim 12, it is rejected for similar reasons as claim 1.
As to Claim 13, it is rejected for similar reasons as claim 2.
As to Claim 14, it is rejected for similar reasons as claim 3.
As to Claim 19, it is rejected for similar reasons as claim 8.
As to Claim 20, it is rejected for similar reasons as claim 1 and 12.
B.
Claims 4-5 and 15-16 are rejected under 35 U.S.C. §103 as being unpatentable over non-patent literature, Li et al. (“Li”), “Effective Document Labeling with Very Few Seed Words: A Topic Modeling Approach,” published in 2016 in view of .non-patent literature, Meng et al. (“Meng”), “Generating Training Data with Language Models: Towards Zero-Shot Language Understanding,” published in 2022 in further view of non-patent literature, Naeem et al. (“Naeem”), “I2MVFormer: Large Language Model Generated Multi-View Document Supervision for Zero-Shot Image Classification,” published in 2023 in further view of non-patent literature, Liu et al. (“Liu”), “Improving Embedding-based Large-scale Retrieval via Label Enhancement,” published in in 2021.
As to Claim 4, Li, Meng and Naeem teach the elements of claim 3.
Li, Meng and Naeem may not explicitly teach: wherein retrieving the first plurality of candidates from the database of data objects based on similarity to the first modified set of seed data objects comprises:
performing a vector similarity search operation within the embedding space to identify the plurality of candidates from the plurality of embeddings, based on a similarity measure between each of the candidates and a respective seed embedding.
Liu teaches in general concepts related to impriving embedding-based retrieval by utilizing label enhancement techniques that incorporate prior knowledge from dynamic term weighting methonds into contextual embeddings (Liu: Abstract). Specifically, Liu teaches that a cosine distance between two different vectors is used in a BERT style model to determine the output score between the query and document.(Liu: Sec. 3.4, p. 137). This framework is applied to the generated label distributions.
It would have been obvious to a person having ordinary skill in the art at a time before the effective filing date of the application to have modified the Li-Meng-Naeem disclosures and teachings by utilizing a vector similarity search operation for the embedding space as taught and suggested by Liu. Such a person would have been motivated to do so with a reasonable expectation of success to allow for a well-known operation to search embedding space with predictable results and performance.
As to Claim 5, Li, Meng, Naeem and Liu teach the elements of claim 4.
Liu further teaches: wherein the similarity measure is a distance measure (Examiner notes that the cosine similarity is a distance measure).
As to Claim 15, it is rejected for similar reasons as claim 4.
As to Claim 16, it is rejected for similar reasons as claim 5.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JAMES T TSAI whose telephone number is (571)270-3916. The examiner can normally be reached M-F 8-5 Eastern.
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, Viker Lamardo can be reached at 571-270-5871. 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.
/JAMES T TSAI/ Primary Examiner, Art Unit 2147
Prosecution Insights