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 .
Response to Arguments
The 101 rejection is overcome by amendments filed 9/9/2026, related to claimed improvements in training. Some of the 112 rejections are overcome, see below, the amendments did not fix all of the 112 rejections.
With respect to the 102 rejections, remarks filed 9/9/2026 are not persuasive.
Applicant argues, “Adding the sigmod non-linearity to the CNN-Transformer model would not result in the CNN-SAN-Transformer model as described in the claim. Because Kong focuses so much on the CNN-Transformer model for detecting the presence or absence of sound events in an audio clip, Kong does not and does not need to teach the CNN-SAN-Transformer model as described in the claim.” Remarks 9. The attention layers of the Self-Attention Network (SAN) are part of the transformer taught by Kong, Kong sec. 1 “A transformer consists of several attention layers.”
Claim Rejections - 35 USC § 112
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.
Claims 1-5 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), 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. Applicant never describes “calculating at least risk values and the weights thereby minimizing classification risk of learning…”
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
Claims 1-5 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), 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.
Claims 1 and 3 recite “calculating at least risk values and the weights thereby minimizing classification risk of learning…” This is never defined in the specification. Further, the phrase as a whole makes no sense, “a risk in correct answer label learning for a purpose of multi-label classification by use of the DNN model…” This is not a term of art.
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.
Claims 1-5 are rejected under 35 U.S.C. 102(a)(1) as being described by Sound Event Detection of Weakly Labelled Data with CNN-Transformer and Automatic Threshold Optimization by Kong et al.
Kong teaches claims 1 and 3. (Currently Amended) A learning device comprising: processing circuitry configured to:
acquire learning data including data and a correct answer label assigned to the data; and (Kong sec. IV “we introduce how we apply the aforementioned architectures for AT and SED training with weakly labelled data.”)
learn a plurality of weights of a deep neural network (DNN) model for predicting an acoustic event from the learning data, wherein (Kong sec. IV “we introduce how we apply the aforementioned architectures for AT and SED training with weakly labelled data.” Training is learning weights. Kong’s title “Sound Event Detection of Weakly Labelled Data…” Kong’s DNN is a “CNN-Transformer”, see abstract.)
the DNN model comprises a sequence of a feature extraction layer and a prediction layer according to a combination of a convolutional neural network (CNN), a self-attention network (SAN) and a Transformer model, Kong’s DNN is a “CNN-Transformer”, see abstract. Kong sec. 1 “A transformer consists of several attention layers.” Feature extraction and prediction layers are what are in a CNN, and the claimed SAN is taught by the attention layers in the transformer.)
the feature extraction layer comprises a sequence of a first block of layers followed by a second block of layers, the first block further comprises respective layers from an input layer of the DNN model to a predetermined intermediate layer of the feature extraction layer, thereby forming a structure that is identical to a structure of layers from an input layer to an intermediate layer of the CNN, (Kong sec. III(B) “To build the CNN-Transformer system, we first apply a CNN described in Section II on the log mel spectrogram of an audio clip. Convolutional layers in the CNN are used to extract high level features of the input log mel spectrogram. We use the feature maps of the last convolutional layer to obtain embedding vectors along time axis.” The CNN is the first block.) the second block further comprises respective layers from the predetermined intermediate layer to an output layer of the feature extraction layer, thereby forming a structure that is identical to a structure of layers from an intermediate layer to an output layer of the SAN, (Kong sec. III(B) “The embedding can be viewed as x with a shape of the number of time frames by the number of channels. The output of the transformer has a shape of T × dv. A fully connected layer followed by a sigmoid non-linearity is applied on this output to predict the presence probabilities of sound classes over time steps.” The transformer/attention layer is the second block.)
the prediction layer predicts an event from an output of the feature extraction layer, and (Kong sec. III(B) “An aggregation function such as average aggregation can be applied to average out those probabilities along time domain to obtain the audio tagging result.” The tagging is the event.)
the learning of the weights further comprises calculating at least risk values and the weights thereby minimizing classification risk of learning by using the learning data and minimizing a loss according to binary cross entropy. (Kong sec. V “The goal of AT or SED is to minimizesome loss J(Θ)…”
PNG
media_image1.png
76
357
media_image1.png
Greyscale
Kong sec. IV(B) “In training, we calculate the categorical binary cross entropy loss between the clip-level prediction F(X) and the ground truth label of X:”)
Kong teaches claim 2. (Currently Amended) The learning device according to claim 1, comprising: processing circuitry configured to:
extract a part of the learning data as a batch; and (Kong sec. IV(A) “waveform of an audio clip as X. For an X lasting for several seconds, we split it into several segments … In training, a classifier f is trained on the segments.” Segments are batches.)
acquire the batch and calculate a risk of loss in correct answer label learning for a purpose of multi-label classification by use of the DNN model (Kong sec. IV(A) “a classifier f is trained on the segments. The loss function can be written…” Equation 3 is the loss function for segment wise training.) and a Binary Cross Entropy loss function; and (Kong sec. IV(B) “we calculate the categorical binary cross entropy loss between the clip-level prediction F(X) and the ground truth label of X…”)
acquire the risk and update the weight of the DNN model so as to minimize the risk. (The training based on the loss function is an update of weights for the classifier F.)
Kong teaches claim 4. (Currently Amended) A computer-readable non-transitory recording medium storing a computer-executable program that when executed by a processor cause a computer to execute as the learning device according to claim 1. (Kong sec. VII “the CNN-transfomer has the advantage of being computed in parallel.”)
Kong teaches claim 5. (Currently Amended) A computer-readable non-transitory recording medium storing a computer-executable program that when executed by a processor cause a computer to execute as the learning device according to claim 2. (Kong sec. VII “the CNN-transfomer has the advantage of being computed in parallel.”)
Conclusion
THIS ACTION IS MADE FINAL. 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 Austin Hicks whose telephone number is (571)270-3377. The examiner can normally be reached Monday - Thursday 8-4 PST.
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, Mariela Reyes can be reached at (571) 270-1006. 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.
/AUSTIN HICKS/ Primary Examiner, Art Unit 2142
