DETAILED ACTION
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 . This action is in response to an amendment filed on February 6th, 2026. Claims 1-20 are pending in the current application with claims 1, 8, and 16 being amended.
Claim Rejections - 35 USC § 102
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.
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.
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claim(s) 16-20 are rejected under 35 U.S.C. 102(a)(2) as being anticipated by Bui et al. (Herein referred to as Bui) (U.S. Patent Application No. US 20220114476 A1)
Regarding claim 16, Bui teaches a system comprising: a memory component; and a
processing device coupled to the memory component, (“Implementations of the present
disclosure may comprise or utilize a special-purpose or general-purpose computer including
computer hardware, such as, for example, one or more processors and system memory, as
discussed in greater detail below.”, Paragraph 167) the processing device to perform
operations comprising: obtaining a first set of labels generated by a teacher model based on
training data; (“the text sequence labeling system trains the student model based on the
unlabeled data and the additional pseudo labels generated by the teacher model.”, Paragraph
23) (The pseudo labels from the teacher model act as out first set of labels.) generating a
second set of labels by at least replacing a label of the first set of labels with labels generated
by a past student model based on the training data; (“the text sequence labeling system learns
a second set of model parameters at the student model based on the set of ground truth text
sequence labels (e.g., the pseudo labels) generated by the teacher model. Moreover, based on
identifying an improved validation output from the student model, the text sequence labeling
system re-initializes the first teacher model with the second set of learned model parameters.”,
Paragraph 20) (The student model learns the second model parameters and changes the
labeling from the teacher model, (acting as our second set of labels) leading to an improved
validation output. If an improvement is shown, the student model replaces the teacher model.) wherein the past student model includes a set of student model parameters from a prior training epoch that are not updated during the current training epoch (The past student model, which is the teacher model, is fixed in Bui.) and labels generated by a past student model provide contextual semantic noise for training the student model (“In some implementations, the teacher model 108 is a bidirectional LSTM neural network that processes text data from the labeled data set 202 to predict corresponding text sequence labels (e.g., key phrases extracted from the text data).”, Paragraph 48) (Key phrases contain contextual semantic information, with any difference in the label corresponding to noise.) and training a student model based on the second set of labels. (“The text sequence labeling system 106 then replaces the teacher model 108 with the re-trained and newly updated student model 110, as shown in FIG. 2. Indeed, in various implementations, the text sequence labeling system 106 continues to repeat these actions until one or more training conditions are met, as further described below.”, Paragraph 55; See also FIG. 2) (The student model and it’s generated “second set of labels” replaces the teacher model, trains a new student model, based on the second set of labels, until a training condition is met. See FIG. 2 for an illustration detailing the process.)
Regarding claim 17, Bui teaches the system of claim 16, wherein the past student model is updated based on a current version of the student model at an expiration of an update interval. (The text sequence labeling system 106 then replaces the teacher model 108 with the re-trained and newly updated student model.”, Paragraph 55) (Simply put, the time spent before replacing the past student model (teacher model) with the newly updated student model would be considered the “update interval.”)
Regarding claim 18, Bui teaches the system of claim 16, wherein the past student model is generated at an expiration of a warmup interval. (Simply put, the time before the past student model (teacher model) is generated would be considered the “warmup interval.”)
Regarding claim 19, Bui teaches the system of claim 16, wherein the processing device is further configured to perform the operations comprising performing a training iteration by at least: obtaining a third set of labels generated by the teacher model; (“the text sequence labeling system trains the student model based on the unlabeled data and the additional pseudo labels generated by the teacher model… The text sequence labeling system 106 then replaces the teacher model 108 with the re-trained and newly updated student model 110, as shown in FIG. 2.”, Paragraphs 23 and 55) (The pseudo labels from the replaced teacher model act as out third set of labels.) generating a fourth set of labels by at least replacing labels of the third set of labels with labels generated by the past student model; (“the text sequence labeling system learns a second set of model parameters at the student model based on the set of ground truth text sequence labels (e.g., the pseudo labels) generated by the teacher model. Moreover, based on identifying an improved validation output from the student model, the text sequence labeling system re-initializes the first teacher model with the second set of learned model parameters.”, Paragraph 20) (The student model learns the second model parameters and changes the labeling from the teacher model, (acting as our fourth set of labels) leading to an improved validation output. If an improvement is shown, the student model replaces the teacher model.) and training the student model based on the fourth set of labels. (“The text sequence labeling system 106 then replaces the teacher model 108 with the re-trained and newly updated student model 110, as shown in FIG. 2. Indeed, in various implementations, the text sequence labeling system 106 continues to repeat these actions until one or more training conditions are met, as further described below.”, Paragraph 55; See also FIG. 2) (The new student model and it’s generated “fourth set of labels” replaces the teacher model, trains yet another new student model, based on the fourth set of labels, until a training condition is met. See FIG. 2 for an illustration detailing the process.)
Regarding claim 20, Bui teaches the system of claim 16, wherein the labels generated by the past student model contain contextual semantic information. (“In some implementations, the teacher model 108 is a bidirectional LSTM neural network that processes text data from the labeled data set 202 to predict corresponding text sequence labels (e.g., key phrases extracted from the text data).”, Paragraph 48) (Key phrases contain contextual semantic information)
Allowable Subject Matter
Claims 1-15 are allowable over prior art and subject matter eligibility. The steps, “obtaining a first set of labels…”, “providing…”, “causing the student model to generate a second set of labels…”, and “modifying…” are all performed “during a current training epoch” which makes the independent claims 1 and 8 allowable, and the claims dependent on 1 and 8 allowable as well.
Response to Arguments
Applicant's arguments filed on February 6th, 2026 have been fully considered but they are not persuasive. The applicant argues in substance:
Argument 1: The reference from the prior rejection do not teach the new limitation “the past student model includes a set of student model parameters from a prior training epoch that are not updated during the current training epoch and the second label provides contextual semantic noise for training the student model.”
The examiner respectfully disagrees. As explained in this action, the teacher model’s parameters of Bui are fixed. Under the broadest reasonable interpretation, this corresponds to a set of student model parameters from a prior training epoch not updated during the current training epoch. As such, the rejection is proper.
Argument 2: Bui does not teach replacing a first label of the first set of labels with a second label generated by a past student model based on a value exceeding a probability threshold, wherein the past student model includes a set of student model parameters from a prior training epoch that are not updated during the current training epoch and the second label provides contextual semantic noise for training the student model. Specifically, Bui does not teach using a snapshot of the student model parameters from a previous epoch to generate labels that are probabilistically mixed with teacher labels during ongoing training, not using outputs from a prior student model as a source of contextual semantic noise within a single training epoch.
The examiner respectfully disagrees. The applicant asserts that Bui’s approach involves replacing the teacher model wholesale with the current student model after validation, and the examiner agrees. This process does teach, under the broadest reasonable interpretation, using labels, (outputs) generated by a teacher model, (previous student model) as a source of contextual semantic noise, as interpreted broadly, a difference in the labels would be considered noise. If the student model, then outperforms the teacher model, the teacher model is replaced by the student model. This is all done within a single training iteration (epoch), and continues until a condition is met. It should be noted that if the invention is able to “use a snapshot of the student model parameters from a previous epoch to generate labels that are probabilistically mixed with teacher labels”, then it is not accurately reflected within the current claim language, and it is possible that this feature would be allowable if clearly recited in the claims.
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 Tyler E Iles whose telephone number is (571)272-5442. The examiner can normally be reached 9:00am - 5:00pm.
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, Kakali Chaki can be reached at (571) 272-3719. 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.
/T.E.I./Patent Examiner, Art Unit 2122
/KAKALI CHAKI/Supervisory Patent Examiner, Art Unit 2122
Prosecution Insights