DETAILED ACTION
1. This is a Non-Final Office Action Correspondence in response to RCE U.S. Application No. 17/559636 filed on March 11, 2026.
Continued Examination Under 37 CFR 1.114
2. A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114.
Notice of Pre-AIA or AIA Status
3. The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Applicant
4. Applicant is encouraged to the contact the Examiner in hopes of reaching allowable subject matter in light of compact prosecution.
Response to Arguments
5. Applicant’s arguments have been considered but are not fully persuasive.
On Pg. 10-12 of remarks in regards to the 35 U.S.C. 103 rejection, Applicant argues the amendments claims.
Examiner replies that a new reference was introduced to teach the amended limitations.
Claim Rejections - 35 USC § 112
6. The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention.
7. Claims 1 and 15 are rejected under 35 U.S.C. 112(a) first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for applications subject to pre-AIA 35 U.S.C. 112, the inventor(s), at the time the application was filed, had possession of the claimed invention. In particular the claimed invention recites the limitation of “and data that identifies a corresponding location in the compressed index vector of each duplicate index”. However the specification Par. [0037] discloses or producing the record of locations of duplicate input index indices for each part, each node produces a record that identifies: (1) locations of unique input indices in each input index vector in the compressed set of input index vectors for that part, (2) locations of removed input indices that are duplicates of the unique input indices in each input index vector in that part, and (3) a number of indices in the input index vectors in the compressed set of input index vectors for that part.
Examiner recites that the specification only discusses removing duplicate indexes in the compressed index vector.
8. Claims 1 and 15 are rejected under 35 U.S.C. 112(b) second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention. In particular the claimed invention recites the limitation of “and data that identifies a corresponding location in the compressed index vector of each duplicate index”. However, the specification par. 0037 only discuses removing duplicate indexes in the compressed index vector.
Claim Rejections - 35 USC § 103
9. In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
10. 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.
11. Claim(s) 1, 2, 6, 8, 14, 15 and 19 is/are rejected under 35 U.S.C. 103 as being unpatentable over Arora et al. U.S. Patent Application Publication No. 2021/0241343 (herein as ‘Arora’) in combination with Xu et al. U.S. Patent Application Publication No. 2021/0201090 (herein as ‘Xu’) as further in view of Shilane et al. U.S. Patent No. 8,935,446 (herein as ‘Shilane’).
As to claim 1 Arora teaches an electronic device, comprising: a plurality of nodes, each node configured to:
Arora teaches: receive an input index vector (Par. 0069 Xu discloses receiving a vector index containing image data);
comprising a plurality of indices including at least one index that is a duplicate of another index in the index vector (Par. 0088 Arora discloses the item embeddings are vectors. Par. 0092 Arora discloses merging duplicate items into a single item list; therefore, the duplicate will be excluded. Par. 105 Arora discloses a model using items in a basket set and creating a query vector using pre-trained user data and items embeddings);
Arora does not teach but Xu teaches wherein the index vector is for use in retrieving embedding table data associated with a machine learning model (Par. 0037-0039 Xu discloses using a first machine learning model to output labels. The output labels are seen as the embedding table data);
Arora and Xu are analogous art because they are in the same field of endeavor, data processing. It would have been obvious to one of ordinary skill in the art, before the effective filing date, to modify the input vector index of Arora to include the location and removal of duplicate records of Xu, to allow for improved accuracy. The suggestion/motivation to combine is that it would be obvious to try in order to speed up classification during video reviewing (Par. 0002-0003 Xu).
Arora teaches generate a compressed index vector that includes unique indices of the index vector and excludes duplicate indices of the index vector (Par. 0079 Arora discloses creating a set of merged lists of item recommendations. When the same items are encountered, only the item with the high score is included in the merged list, and therefore the lower score item is removed from the merged part. Par. 0088 Arora discloses the item embeddings are vectors. Par. 0092 Arora discloses merging duplicate items into a single item list; therefore, the duplicate will be excluded. Par. 105 Arora discloses a model using items in a basket set and creating a query vector using pre-trained user data and items embeddings. The multiple parts are seen as the item embeddings. Arora discloses concatenating the item embeddings. Concatenating the item embeddings is seen as compressing lookup data. Arora discloses creating and index using the items. Arora discloses the items embeddings can be stored for fast lookup. The item embeddings based upon the item selected by the user is seen as the input index vector);
Arora does not teach but Shilane teaches generate location data comprising: data that identifies each duplicate index in the index vector (Col. 9 Lines 26-36 Shilane discloses duplicate entries that point to the particular location of the base entry. The duplicate entry is seen as the duplicate index);
Arora and Shilane are analogous art because they are in the same field of endeavor, data processing. It would have been obvious to one of ordinary skill in the art, before the effective filing date, to modify the input vector index of Arora to include the location and removal of duplicate records of Shilane, to allow reduce the search time. The suggestion/motivation to combine is that it would be obvious to try in order to reduce the time and resources needed to search large number of records (Col. 1 Lines 20-38 Shilane).
Arora teaches and data that identifies a corresponding location in the compressed index vector of each duplicate index (Par. 0092 Arora discloses merges duplicate items into a single item list, therefore the duplicate will not be included);
and communicate, to one or more other computing nodes of the plurality of computing nodes, the compressed index vector and the location data for use in processing by the machine learning model (Par. 0048 Arora discloses training a model based upon millions of users and storing the model. Par.0075 Arora discloses receiving the trained item embeddings to generate recommendations. Par. 0083 Arora discloses communicating the training model to many different users to use in updated retraining based upon the past purchase history of the users. Communicating the model to the users is seen as communicating the index. Using the item beddings as a search to the top 30 nearest neighbors is seen as communicating the compressed lookup data for a respective part to each other node).
As to claim 2 Arora in combination with Shilane and Xu teaches each and every limitation of claim 1.
Arora teaches wherein each node is further configured to: generate compressed lookup data based at least in part on the generated compressed input index vectors (Par. 0079 Arora discloses creating a set of merged list of item recommendations. When the same items is encountered, only the item with the high score is included in the merged list, and therefore the lower score item is removed from the merged part);
and-communicate, to the one or more other computing nodes of the plurality of computing nodes, the compressed lookup data after communicating the location data (Par. 105 Arora discloses performing a batch lookup for each item in the basket sets of the top 30 nearest neighbors. Performing the search is seen as using the lookup of the item beddings to discover if they exist within the storage of each neighbor. Using the item beddings as a search to the top 30 nearest neighbors is seen as communicating the compressed lookup data for a respective part to each other node).
As to claim 6 Arora in combination with Shilane and Xu teaches each and every limitation of claim 1.
Shilane teaches wherein the location data is configured to enable lossless reconstruction of the index vector at the one or more other computing nodes by associating multiple positions in the index vector with a same index in the compressed index vector (Col. 9 Lines 26-36 Shilane discloses duplicate entries that point to the particular location of the base entry. The duplicate entry is seen as the duplicate index).
As to claim 8 Arora teaches a comprising:
receiving, by processing circuitry of a given computing node, an input index vector comprising a plurality of indices including at least one index that is a duplicate of another index in the index vector (Par. 0088 Arora discloses the item embeddings are vectors. Par. 0092 Arora discloses merging duplicate items into a single item list; therefore, the duplicate will be excluded. Par. 105 Arora discloses a model using items in a basket set and creating a query vector using pre-trained user data and items embeddings);
Arora does not teach but Xu teaches wherein the index vector is for use in retrieving embedding table data associated with a machine learning mode (Par. 0037-0039 Xu discloses using a first machine learning model to output labels. The output labels are seen as the embedding table data);
Arora and Xu are analogous art because they are in the same field of endeavor, data processing. It would have been obvious to one of ordinary skill in the art, before the effective filing date, to modify the input vector index of Arora to include the location and removal of duplicate records of Xu, to allow for improved accuracy. The suggestion/motivation to combine is that it would be obvious to try in order to speed up classification during video reviewing (Par. 0002-0003 Xu).
Arora teaches generating, by the processing circuitry, a compressed input index vector that includes unique input indices of the input index vector and excludes duplicate input indices of the input index vector (Par. 0079 Arora discloses creating a set of merged lists of item recommendations. When the same items are encountered, only the item with the high score is included in the merged list, and therefore the lower score item is removed from the merged part. Par. 0088 Arora discloses the item embeddings are vectors. Par. 0092 Arora discloses merging duplicate items into a single item list; therefore, the duplicate will be excluded. Par. 105 Arora discloses a model using items in a basket set and creating a query vector using pre-trained user data and items embeddings. The multiple parts are seen as the item embeddings. Arora discloses concatenating the item embeddings. Concatenating the item embeddings is seen as compressing lookup data. Arora discloses creating and index using the items. Arora discloses the items embeddings can be stored for fast lookup. The item embeddings based upon the item selected by the user is seen as the input index vector);
Arora does not teach but Shilane teaches generating, by the processing circuitry, location data comprising: data that identifies each duplicate index in the input index vector (Col. 9 Lines 26-36 Shilane discloses duplicate entries that point to the particular location of the base entry. The duplicate entry is seen as the duplicate index);
Arora and Shilane are analogous art because they are in the same field of endeavor, data processing. It would have been obvious to one of ordinary skill in the art, before the effective filing date, to modify the input vector index of Arora to include the location and removal of duplicate records of Shilane, to allow reduce the search time. The suggestion/motivation to combine is that it would be obvious to try in order to reduce the time and resources needed to search large number of records (Col. 1 Lines 20-38 Shilane).
Arora teaches and data that identifies a corresponding location in the compressed index vector of each duplicate index (Par. 0092 Arora discloses merges duplicate items into a single item list, therefore the duplicate will not be included);
and communicating, by the processing circuitry, to one or more other computing nodes of a plurality of computing nodes, the compressed input index vector and the location data for use in processing by the machine learning mode (Par. 0048 Arora discloses training a model based upon millions of users and storing the model. Par.0075 Arora discloses receiving the trained item embeddings to generate recommendations. Par. 0083 Arora discloses communicating the training model to many different users to use in updated retraining based upon the past purchase history of the users. Communicating the model to the users is seen as communicating the index. Using the item beddings as a search to the top 30 nearest neighbors is seen as communicating the compressed lookup data for a respective part to each other node).
As to claim 14 Arora in combination with Shilane and Xu teaches each and every limitation of claim 8.
Arora teaches wherein the location data is configured to enable lossless reconstruction of the index vector at the one or more other computing nodes by associating multiple positions in the index vector with a same index in the compressed index vector (Col. 9 Lines 26-36 Shilane discloses duplicate entries that point to the particular location of the base entry. The duplicate entry is seen as the duplicate index).
As to claim 15 Arora teaches a computing node, comprising:
at least one local memory storing a portion of embedding table data of a machine learning model; and at least one processor comprising circuitry configured to:
a plurality of nodes, each node configured to (Par. 0028 Arora discloses servers. The severs are seen as nodes):
receive an input index vector comprising a plurality of indices including at least one index that is a duplicate of another index in the index vector (Par. 0084 Arora discloses receiving items selected by a user. Par. 0088 Arora discloses the items are associated with vectors. Par. 0088 Arora discloses the item embeddings are vectors. Par. 0092 Arora discloses merging duplicate items into a single item list; therefore, the duplicate will be excluded. Par. 105 Arora discloses a model using items in a basket set and creating a query vector using pre-trained user data and items embeddings);
Arora does not teach but Xu teaches wherein the index vector is for use in retrieving embedding table data associated with a machine learning model (Par. 0037-0039 Xu discloses using a first machine learning model to output labels. The output labels are seen as the embedding table data);
Arora and Xu are analogous art because they are in the same field of endeavor, data processing. It would have been obvious to one of ordinary skill in the art, before the effective filing date, to modify the input vector index of Arora to include the location and removal of duplicate records of Xu, to allow for improved accuracy. The suggestion/motivation to combine is that it would be obvious to try in order to speed up classification during video reviewing (Par. 0002-0003 Xu).
Arora teaches generate a compressed index vector comprising each unique index of the and excludes duplicate indices of the index vector (Par. 0079 Arora discloses creating a set of merged lists of item recommendations. When the same items are encountered, only the item with the high score is included in the merged list, and therefore the lower score item is removed from the merged part. Par. 0088 Arora discloses the item embeddings are vectors. Par. 0092 Arora discloses merging duplicate items into a single item list; therefore, the duplicate will be excluded. Par. 105 Arora discloses a model using items in a basket set and creating a query vector using pre-trained user data and items embeddings. The multiple parts are seen as the item embeddings. Arora discloses concatenating the item embeddings. Concatenating the item embeddings is seen as compressing lookup data. Arora discloses creating and index using the items. Arora discloses the items embeddings can be stored for fast lookup. The item embeddings based upon the item selected by the user is seen as the input index vector);
Arora does not teach but Shilane teaches generate location data comprising: data that identifies each duplicate index in the index vector and data that identifies corresponding location in the compressed index vector (Col. 9 Lines 26-36 Shilane discloses duplicate entries that point to the particular location of the base entry. The duplicate entry is seen as the duplicate index);
Arora and Shilane are analogous art because they are in the same field of endeavor, data processing. It would have been obvious to one of ordinary skill in the art, before the effective filing date, to modify the input vector index of Arora to include the location and removal of duplicate records of Shilane, to allow reduce the search time. The suggestion/motivation to combine is that it would be obvious to try in order to reduce the time and resources needed to search large number of records (Col. 1 Lines 20-38 Shilane).
Xu teaches retrieve lookup data from the portion of the embedding table data stored in the local memory using the compressed index vector (Par. 0037-0039 Xu discloses using a first machine learning model to output labels. The output labels are seen as the embedding table data);
Arora teaches and communicate, to one or more other computing nodes, the compressed input index vector and the location data for use in processing by the machine learning model (Par. 0048 Arora discloses training a model based upon millions of users and storing the model. Par.0075 Arora discloses receiving the trained item embeddings to generate recommendations. Par. 0083 Arora discloses communicating the training model to many different users to use in updated retraining based upon the past purchase history of the users. Communicating the model to the users is seen as communicating the index. Using the item beddings as a search to the top 30 nearest neighbors is seen as communicating the compressed lookup data for a respective part to each other node).
As to claim 19 Arora in combination with Shilane and Xu teaches each and every limitation of claim 15.
Shilane teaches wherein the location data is configured to enable lossless reconstruction of the index vector at the one or more other computing nodes by associating multiple positions in the index vector with a same index in the compressed index vector (Col. 9 Lines 26-36 Shilane discloses duplicate entries that point to the particular location of the base entry. The duplicate entry is seen as the duplicate index).
9. Claim(s) 3-7, 9-13, 16-18 and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Arora et al. U.S. Patent Application Publication No. 2021/0241343 (herein as ‘Arora’) in combination with Xu et al. U.S. Patent Application Publication No. 2021/0201090 (herein as ‘Xu’), Shilane et al. U.S. Patent No. 8,935,446 (herein as ‘Shilane’) and further in view of Khatibi et al. U.S. Patent Application Publication No. 2022/0222543 (herein as ‘Khatibi’).
As to claim 3 Arora in combination with Shilane and Xu teaches each and every limitation of claim 2.
Arora does not teach but Khatibi teaches wherein the processing circuitry is further configured to: produce a record of locations that identifies: locations of the unique input indices in the compressed input index vector (Pp.77 Khatibi discloses using the information to locate duplicate the data sets);
locations of the duplicate input indices in the compressed input index vector; and a number of input indices in the compressed input index vectors; and generate the location data based at least in part on the record of locations (Par. 0093 Khatibi discloses identifying a number of duplicate text. Par. 0077 Khatibi discloses merging duplicate records. Merging duplicate records is seen as removing an input).
Arora and Khatibi are analogous art because they are in the same field of endeavor, data processing. It would have been obvious to one of ordinary skill in the art, before the effective filing date, to modify the input vector index of Arora to include the location and removal of duplicate records of Khatibi, to allow reduce the search time. The suggestion/motivation to combine is that it would be obvious to try in order to reduce the time and resources needed to search large number of records (Par. 0003 Khatibi).
As to claim 4 Arora in combination with Shilane and Xu teaches each and every limitation of claim 1.
Arora does not teach but Khatibi teaches wherein the processing circuitry is configured to generate the compressed input index vector and the location data for each part among multiple parts of the data to be processed by the machine learning model, the multiple parts corresponding to partitions of the data for parallel processing across the plurality of computing nodes (Pp.77 Khatibi discloses using the information to locate duplicate the data sets. Par. 0077 Khatibi discloses merging duplicate records. Merging duplicate records is seen as removing an input).
Arora and Khatibi are analogous art because they are in the same field of endeavor, data processing. It would have been obvious to one of ordinary skill in the art, before the effective filing date, to modify the input vector index of Arora to include the location and removal of duplicate records of Khatibi, to allow reduce the search time. The suggestion/motivation to combine is that it would be obvious to try in order to reduce the time and resources needed to search large number of records (Par. 0003 Khatibi).
As to claim 5 Arora in combination with Shilane and Xu teaches each and every limitation of claim 1.
Arora does not teach but Khatibi teaches wherein the embedding table data of the machine learning model is distributed across local memories of the plurality of computing nodes such that each computing node stores only a portion of the embedding table data, and wherein the compressed index vector and the location data are communicated to enable retrieval of embedding table data from portions stored in other computing nodes (Par. 0038 Arora discloses using a machine learning on one or more databases to obtain collection data).
Arora and Khatibi are analogous art because they are in the same field of endeavor, data processing. It would have been obvious to one of ordinary skill in the art, before the effective filing date, to modify the input vector index of Arora to include the location and removal of duplicate records of Khatibi, to allow reduce the search time. The suggestion/motivation to combine is that it would be obvious to try in order to reduce the time and resources needed to search large number of records (Par. 0003 Khatibi).
As to claim 7 Arora in combination with Shilane and Xu teaches each and every limitation of claim 1.
In addition Khatibi teaches wherein the processing circuitry is further configured to: acquires lookup data from at least one embedding table stored in a local memory using the compressed input index vectors, the lookup data including data from rows of the at least one embedding table, and the compressed input index vector comprising indices identifying rows of the at least one embedding table (Pp.77 Khatibi discloses using the information to locate duplicate the data sets).
Arora and Khatibi are analogous art because they are in the same field of endeavor, data processing. It would have been obvious to one of ordinary skill in the art, before the effective filing date, to modify the input vector index of Arora to include the location and removal of duplicate records of Khatibi, to allow reduce the search time. The suggestion/motivation to combine is that it would be obvious to try in order to reduce the time and resources needed to search large number of records (Par. 0003 Khatibi).
As to claim 9 Arora in combination with Shilane and Xu teaches each and every limitation of claim 8.
Arora teaches further comprising, generating, by the processing circuitry, compressed lookup data based at least in part on the compressed: input index vectors and communicating, by the processing circuitry, to the one or more other computing nodes of the plurality of computing nodes, the compressed lookup data after communicating the location data (Par. 0079 Arora discloses creating a set of merged lists of item recommendations. When the same items is encountered, only the item with the high score is included in the merged list, and therefore the lower score item is removed from the merged part).
As to claim 10 Arora in combination with Shilane and Xu teaches each and every limitation of claim 9.
In addition Khatibi teaches further comprising, producing, by the processing circuity, a record of locations that identifies: locations of the unique input indices in the compressed input index vector (Pp.77 Khatibi discloses using the information to locate duplicate the data sets);
locations of the duplicate input indices in the compressed input index vector; and a number of input indices in the compressed input index vectors; and generating, by the processing circuity the location data based at least in part on the record of locations (Par. 0077 Khatibi discloses merging duplicate records. Merging duplicate records is seen as removing an input).
Arora and Khatibi are analogous art because they are in the same field of endeavor, data processing. It would have been obvious to one of ordinary skill in the art, before the effective filing date, to modify the input vector index of Arora to include the location and removal of duplicate records of Khatibi, to allow reduce the search time. The suggestion/motivation to combine is that it would be obvious to try in order to reduce the time and resources needed to search large number of records (Par. 0003 Khatibi).
As to claim 11 Arora in combination with Shilane and Xu teaches each and every limitation of claim 9.
In addition Khatibi teaches further comprising generating, by the processing circuitry, the compressed input index vector and the location data for each part among multiple parts of the data to be processed by the machine learning model, the multiple parts corresponding to partitions of the data for parallel processing across the plurality of computing nodes (Pp.77 Khatibi discloses using the information to locate duplicate the data sets. Par. 0077 Khatibi discloses merging duplicate records. Merging duplicate records is seen as removing an input).
As to claim 12 Arora in combination with Shilane and Xu teaches each and every limitation of claim 8.
In addition Khatibi teaches further comprising acquiring by the processing circuitry, lookup data from at least one embedding table stored in a local memory using the compressed input index vectors, the lookup data comprising data from rows of the at least one embedding table, and the compressed input index vector comprising indices identifying rows of the at least one embedding table (Pp.77 Khatibi discloses using the information to locate duplicate the data sets).
As to claim 13 Arora in combination with Shilane and Xu teaches each and every limitation of claim 8.
In addition Khatibi teaches wherein processing data compressed input index vector and the location data by the machine learning model occurs during one or more of: training operations for training the machine learning model; or operations for using the machine learning model after the model has been trained (Par. 0107 Khatibi discloses using a machine learning model for clustering).
As to claim 16 Arora in combination with Shilane and Xu teaches each and every limitation of claim 15.
In addition Khatibi discloses wherein the circuitry is further configured to generate the compressed index vector by removing duplicate indices prior to retrieving embedding table data, such that each unique index in the compressed index vector is used for at most one embedding table lookup (Par. 0059 Khatibi discloses removing duplicate copies).
As to claim 17 Arora in combination with Shilane and Xu teaches each and every limitation of claim 15.
In addition Khatibi teaches generate compressed lookup data based at least in part on the generated compressed input index vector; and communicate, to the one or more other computing nodes of the plurality of computing nodes, the compressed lookup data after communicating the location data (Par. 105 Arora discloses a model using items in a basket set and creating a query vector using pre-trained user data and items embeddings. The multiple parts are seen as the item embeddings. Arora disclose concatenating the item embeddings. Arora discloses creating and index using the items. Arora discloses the items embeddings can be stored for fast lookup. The fast look up is seen as decompressed lookup data. The index of query vectors is seen as the input index vector).
Arora and Khatibi are analogous art because they are in the same field of endeavor, data processing. It would have been obvious to one of ordinary skill in the art, before the effective filing date, to modify the input vector index of Arora to include the location and removal of duplicate records of Khatibi, to allow reduce the search time. The suggestion/motivation to combine is that it would be obvious to try in order to reduce the time and resources needed to search large number of records (Par. 0003 Khatibi).
As to claim 18 Arora in combination with Shilane and Xu teaches each and every limitation of claim 17.
In addition wherein the at least one processor is further configured to:produce a record of locations that identifies: locations of the unique input indices in the compressed input index vector (Pp.77 Khatibi discloses using the information to locate duplicate the data sets);
locations of the duplicate input indices in the compressed input index vector and a number of input indices in the compressed input index vector; and generate the location data based at least in part on the record of locations (Par. 0093 Khatibi discloses identifying a number of duplicate text. Par. 0077 Khatibi discloses merging duplicate records. Merging duplicate records is seen as removing an input);
As to claim 20 Arora in combination with Shilane and Xu teaches each and every limitation of claim 15.
Khatibi teaches wherein processing of the compressed input index vector and the location data by the machine learning model occurs during one or more of: training operations for training the machine learning model; or operations for using the machine learning model after the model has been trained (Par. 0107 Khatibi discloses using a machine learning model for clustering).
Conclusion
12. Any inquiry concerning this communication or earlier communications from the examiner should be directed to JERMAINE A MINCEY whose telephone number is (571)270-5010. The examiner can normally be reached 8am EST until 5pm 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, Ann J Lo can be reached on (571) 272-9767. 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.A.M/ March 31, 2026Examiner, Art Unit 2159
/ANN J LO/Supervisory Patent Examiner, Art Unit 2159
Prosecution Insights