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 .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on November 6, 2023 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Claim Objections
Claim 6 is objected to because of the following informalities: “selecting meta-training batch subset” should read “selecting a meta-training batch subset”. Appropriate correction is required.
Claim Rejections - 35 USC § 112
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.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 7 and 17 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.
The term “better accommodate” in claims 7 and 17 is a relative term which renders the claim indefinite. The term “better accommodate” is not defined by the claim, the specification does not provide a standard for ascertaining the requisite degree, and one of ordinary skill in the art would not be reasonably apprised of the scope of the invention. The term “accommodate” has been rendered indefinite by the use of the term “better”.
Examiner’s Note: For the purposes of Examination, the term “better accommodate” will be interpreted as any improvement in classification in domain adaptation including a first and second domain.
Claims 8 and 18 are further rejected for dependence on claims 7 and 17, respectively.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Regarding Claim 1,
Claim 1 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 1 is directed to a computer-implemented method for training a model, which is directed to a process, one of the statutory categories.
Step 2A Prong One Analysis: The limitations:
“determining a dropout mask based on gradient signal to noise ratio (GSNR) of parameters of a neural network model”
“iteratively updating the dropout mask”
As drafted, under their broadest reasonable interpretations, cover mental processes, i.e., concepts performed in the human mind (including an observation, evaluation, judgement, opinion). The above limitations in the context of this claim correspond to mental processes, e.g., evaluation and judgement with assistance of pen and paper.
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application. In particular, the claim recited additional elements that are mere instructions to apply an exception (See MPEP 2106.05(f)).
The limitations:
“A computer-implemented method for training a model”
“training the neural network model with parameters zeroed-out according to the dropout mask”
“performing the training based on the updated dropout mask”
As drafted, are additional elements that amount to no more than mere instructions to apply an exception for the abstract ideas. See MPEP 2106.05(f).
Therefore, the additional elements do not integrate the abstract ideas into a practical application.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to the integration of the abstract ideas into a practical application, all of the additional elements are “mere instructions to apply”. Mere instructions to apply an exception cannot provide an inventive concept. The claim is not patent eligible.
Regarding Claim 2,
Claim 2 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 2 is directed to a computer-implemented method for training a model, which is directed to a process, one of the statutory categories.
Step 2A Prong One Analysis: The limitations:
“determining a dropout ratio that determines a number of parameters to zero-out”
As drafted, under their broadest reasonable interpretations, cover mental processes, i.e., concepts performed in the human mind (including an observation, evaluation, judgement, opinion). The above limitations in the context of this claim correspond to mental processes, e.g., evaluation and judgement with assistance of pen and paper.
Step 2A Prong Two Analysis: See corresponding analysis of claim 1.
Step 2B Analysis: See corresponding analysis of claim 1.
Regarding Claim 3,
Claim 3 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 3 is directed to a computer-implemented method for training a model, which is directed to a process, one of the statutory categories.
Step 2A Prong One Analysis: See corresponding analysis of claim 2.
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application. In particular, the claim recited additional elements that do not apply the exception in a meaningful way (See MPEP 2106.05(e)).
The limitations:
“wherein the dropout ratio varies for different parts of the neural network model”
As drafted, is an additional element that does not apply an exception for the abstract ideas in a meaningful way. See MPEP 2106.05(e).
Therefore, the additional elements do not integrate the abstract ideas into a practical application.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to the integration of the abstract ideas into a practical application, all of the additional elements do not apply the exception in a meaningful way. The claim is not patent eligible.
Regarding Claim 4,
Claim 4 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 4 is directed to a computer-implemented method for training a model, which is directed to a process, one of the statutory categories.
Step 2A Prong One Analysis: See corresponding analysis of claim 1.
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application. In particular, the claim recited additional elements that are mere instructions to apply an exception (See MPEP 2106.05(f)).
The limitations:
“performing meta-training and meta-testing to update a loss function”
As drafted, are additional elements that amount to no more than mere instructions to apply an exception for the abstract ideas. See MPEP 2106.05(f).
Therefore, the additional elements do not integrate the abstract ideas into a practical application.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to the integration of the abstract ideas into a practical application, all of the additional elements are “mere instructions to apply”. Mere instructions to apply an exception cannot provide an inventive concept. The claim is not patent eligible.
Regarding Claim 5,
Claim 5 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 5 is directed to a computer-implemented method for training a model, which is directed to a process, one of the statutory categories.
Step 2A Prong One Analysis: The limitations:
“determining a gradient of the loss function”
As drafted, under their broadest reasonable interpretations, cover mental processes, i.e., concepts performed in the human mind (including an observation, evaluation, judgement, opinion). The above limitations in the context of this claim correspond to mental processes, e.g., evaluation and judgement with assistance of pen and paper.
Step 2A Prong Two Analysis: See corresponding analysis of claim 3.
Step 2B Analysis: See corresponding analysis of claim 3.
Regarding Claim 6,
Claim 6 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 6 is directed to a computer-implemented method for training a model, which is directed to a process, one of the statutory categories.
Step 2A Prong One Analysis: The limitations:
“selecting meta-training batch subset and a meta-testing batch subset, with examples of the meta-testing batch subset being selected according to their distance from the meta-training batch subset”
As drafted, under their broadest reasonable interpretations, cover mental processes, i.e., concepts performed in the human mind (including an observation, evaluation, judgement, opinion). The above limitations in the context of this claim correspond to mental processes, e.g., evaluation and judgement with assistance of pen and paper.
Step 2A Prong Two Analysis: See corresponding analysis of claim 3.
Step 2B Analysis: See corresponding analysis of claim 3.
Regarding Claim 7,
Claim 7 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 7 is directed to a computer-implemented method for training a model, which is directed to a process, one of the statutory categories.
Step 2A Prong One Analysis: See corresponding analysis of claim 1.
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application. In particular, the claim recited additional elements that are mere instructions to apply an exception (See MPEP 2106.05(f)).
The limitations:
“wherein training the neural network model includes a training dataset of examples in a first domain and wherein the dropout mask causes the training to better accommodate testing examples from second domains that are not included in the training dataset”
As drafted, are additional elements that amount to no more than mere instructions to apply an exception for the abstract ideas. See MPEP 2106.05(f).
Therefore, the additional elements do not integrate the abstract ideas into a practical application.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to the integration of the abstract ideas into a practical application, all of the additional elements are “mere instructions to apply”. Mere instructions to apply an exception cannot provide an inventive concept. The claim is not patent eligible.
Regarding Claim 8,
Claim 8 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 8 is directed to a computer-implemented method for training a model, which is directed to a process, one of the statutory categories.
Step 2A Prong One Analysis: See corresponding analysis of claim 7.
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application. In particular, the claim recited additional elements that do not apply the exception in a meaningful way (See MPEP 2106.05(e)).
The limitations:
“wherein the training dataset includes images and wherein the first domain and the second domains differ according to environmental conditions or geographic location”
As drafted, is an additional element that does not apply an exception for the abstract ideas in a meaningful way. See MPEP 2106.05(e).
Therefore, the additional elements do not integrate the abstract ideas into a practical application.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to the integration of the abstract ideas into a practical application, all of the additional elements do not apply the exception in a meaningful way. The claim is not patent eligible.
Regarding Claim 9,
Claim 9 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 9 is directed to a computer-implemented method for training a model, which is directed to a process, one of the statutory categories.
Step 2A Prong One Analysis: See corresponding analysis of claim 1.
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application. In particular, the claim recited additional elements that are mere instructions to apply an exception (See MPEP 2106.05(f)).
The limitations:
“wherein training the neural network model with parameters zeroed-out includes performing a feed-forward operation where parameters designated by the dropout mask are omitted”
As drafted, are additional elements that amount to no more than mere instructions to apply an exception for the abstract ideas. See MPEP 2106.05(f).
Therefore, the additional elements do not integrate the abstract ideas into a practical application.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to the integration of the abstract ideas into a practical application, all of the additional elements are “mere instructions to apply”. Mere instructions to apply an exception cannot provide an inventive concept. The claim is not patent eligible.
Regarding Claim 10,
Claim 10 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 10 is directed to a computer-implemented method for training a model, which is directed to a process, one of the statutory categories.
Step 2A Prong One Analysis: See corresponding analysis of claim 1.
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application. In particular, the claim recited additional elements that do not apply the exception in a meaningful way (See MPEP 2106.05(e)).
The limitations:
“wherein the GSNR of a parameter is determined as a ratio between the parameter’s mean gradients with respect to a loss function and a corresponding variance”
As drafted, is an additional element that does not apply an exception for the abstract ideas in a meaningful way. See MPEP 2106.05(e).
Therefore, the additional elements do not integrate the abstract ideas into a practical application.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to the integration of the abstract ideas into a practical application, all of the additional elements do not apply the exception in a meaningful way. The claim is not patent eligible.
Regarding Claim 11,
Claim 11 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 11 is directed to a system for training a model, which is directed to a machine, one of the statutory categories.
Step 2A Prong One Analysis: The limitations:
“determine a dropout mask based on gradient signal to noise ratio (GSNR) of parameters of a neural network model”
“iteratively update the dropout mask”
As drafted, under their broadest reasonable interpretations, cover mental processes, i.e., concepts performed in the human mind (including an observation, evaluation, judgement, opinion). The above limitations in the context of this claim correspond to mental processes, e.g., evaluation and judgement with assistance of pen and paper.
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application. In particular, the claim recited additional elements that are mere instructions to apply an exception (See MPEP 2106.05(f)).
The limitations:
“A system for training a model, comprising: a hardware processor; and a memory that stores a computer program which, when executed by the hardware processor, causes the hardware processor to”
“train the neural network model with parameters zeroed-out according to the dropout mask”
“performing the training based on the updated dropout mask”
As drafted, are additional elements that amount to no more than mere instructions to apply an exception for the abstract ideas. See MPEP 2106.05(f).
Therefore, the additional elements do not integrate the abstract ideas into a practical application.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to the integration of the abstract ideas into a practical application, all of the additional elements are “mere instructions to apply”. Mere instructions to apply an exception cannot provide an inventive concept. The claim is not patent eligible.
Regarding Claim 12,
Claim 12 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 12 is directed to a system for training a model, which is directed to a machine, one of the statutory categories.
Step 2A Prong One Analysis: The limitations:
“determine a dropout ratio that determines a number of parameters to zero-out”
As drafted, under their broadest reasonable interpretations, cover mental processes, i.e., concepts performed in the human mind (including an observation, evaluation, judgement, opinion). The above limitations in the context of this claim correspond to mental processes, e.g., evaluation and judgement with assistance of pen and paper.
Step 2A Prong Two Analysis: See corresponding analysis of claim 11.
Step 2B Analysis: See corresponding analysis of claim 11.
Regarding Claim 13,
Claim 13 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 13 is directed to a system for training a model, which is directed to a machine, one of the statutory categories.
Step 2A Prong One Analysis: See corresponding analysis of claim 12.
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application. In particular, the claim recited additional elements that do not apply the exception in a meaningful way (See MPEP 2106.05(e)).
The limitations:
“wherein the dropout ratio varies for different parts of the neural network model”
As drafted, is an additional element that does not apply an exception for the abstract ideas in a meaningful way. See MPEP 2106.05(e).
Therefore, the additional elements do not integrate the abstract ideas into a practical application.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to the integration of the abstract ideas into a practical application, all of the additional elements do not apply the exception in a meaningful way. The claim is not patent eligible.
Regarding Claim 14,
Claim 14 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 14 is directed to a system for training a model, which is directed to a machine, one of the statutory categories.
Step 2A Prong One Analysis: See corresponding analysis of claim 11.
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application. In particular, the claim recited additional elements that are mere instructions to apply an exception (See MPEP 2106.05(f)).
The limitations:
“performing meta-training and meta-testing to update a loss function to determine the dropout mask”
As drafted, are additional elements that amount to no more than mere instructions to apply an exception for the abstract ideas. See MPEP 2106.05(f).
Therefore, the additional elements do not integrate the abstract ideas into a practical application.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to the integration of the abstract ideas into a practical application, all of the additional elements are “mere instructions to apply”. Mere instructions to apply an exception cannot provide an inventive concept. The claim is not patent eligible.
Regarding Claim 15,
Claim 15 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 15 is directed to a system for training a model, which is directed to a machine, one of the statutory categories.
Step 2A Prong One Analysis: The limitations:
“determine a gradient of the loss function to determine the dropout mask”
As drafted, under their broadest reasonable interpretations, cover mental processes, i.e., concepts performed in the human mind (including an observation, evaluation, judgement, opinion). The above limitations in the context of this claim correspond to mental processes, e.g., evaluation and judgement with assistance of pen and paper.
Step 2A Prong Two Analysis: See corresponding analysis of claim 14.
Step 2B Analysis: See corresponding analysis of claim 14.
Regarding Claim 16,
Claim 16 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 16 is directed to a system for training a model, which is directed to a machine, one of the statutory categories.
Step 2A Prong One Analysis: The limitations:
“select a meta-training batch subset and a meta-testing batch subset, with examples of the meta-testing batch subset being selected according to their distance from the meta-training batch subset”
As drafted, under their broadest reasonable interpretations, cover mental processes, i.e., concepts performed in the human mind (including an observation, evaluation, judgement, opinion). The above limitations in the context of this claim correspond to mental processes, e.g., evaluation and judgement with assistance of pen and paper.
Step 2A Prong Two Analysis: See corresponding analysis of claim 14.
Step 2B Analysis: See corresponding analysis of claim 14.
Regarding Claim 17,
Claim 17 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 17 is directed to a system for training a model, which is directed to a machine, one of the statutory categories.
Step 2A Prong One Analysis: See corresponding analysis of claim 11.
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application. In particular, the claim recited additional elements that are mere instructions to apply an exception (See MPEP 2106.05(f)).
The limitations:
“wherein the computer program further causes the hardware processor to use a training dataset of examples in a first domain and wherein the dropout mask causes the training to better accommodate testing examples from second domains that are not included in the training dataset”
As drafted, are additional elements that amount to no more than mere instructions to apply an exception for the abstract ideas. See MPEP 2106.05(f).
Therefore, the additional elements do not integrate the abstract ideas into a practical application.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to the integration of the abstract ideas into a practical application, all of the additional elements are “mere instructions to apply”. Mere instructions to apply an exception cannot provide an inventive concept. The claim is not patent eligible.
Regarding Claim 18,
Claim 18 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 18 is directed to a system for training a model, which is directed to a machine, one of the statutory categories.
Step 2A Prong One Analysis: See corresponding analysis of claim 17.
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application. In particular, the claim recited additional elements that do not apply the exception in a meaningful way (See MPEP 2106.05(e)).
The limitations:
“wherein the training dataset includes images and wherein the first domain and the second domains differ according to environmental conditions or geographic location”
As drafted, is an additional element that does not apply an exception for the abstract ideas in a meaningful way. See MPEP 2106.05(e).
Therefore, the additional elements do not integrate the abstract ideas into a practical application.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to the integration of the abstract ideas into a practical application, all of the additional elements do not apply the exception in a meaningful way. The claim is not patent eligible.
Regarding Claim 19,
Claim 19 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 19 is directed to a system for training a model, which is directed to a machine, one of the statutory categories.
Step 2A Prong One Analysis: See corresponding analysis of claim 11.
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application. In particular, the claim recited additional elements that are mere instructions to apply an exception (See MPEP 2106.05(f)).
The limitations:
“wherein the computer program further causes the hardware processor to perform a feed-forward operation where parameters designated by the dropout mask are omitted”
As drafted, are additional elements that amount to no more than mere instructions to apply an exception for the abstract ideas. See MPEP 2106.05(f).
Therefore, the additional elements do not integrate the abstract ideas into a practical application.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to the integration of the abstract ideas into a practical application, all of the additional elements are “mere instructions to apply”. Mere instructions to apply an exception cannot provide an inventive concept. The claim is not patent eligible.
Regarding Claim 20,
Claim 20 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1 Analysis: Claim 20 is directed to a system for training a model, which is directed to a machine, one of the statutory categories.
Step 2A Prong One Analysis: See corresponding analysis of claim 11.
Step 2A Prong Two Analysis: The judicial exceptions are not integrated into a practical application. In particular, the claim recited additional elements that do not apply the exception in a meaningful way (See MPEP 2106.05(e)).
The limitations:
“wherein the GSNR of a parameter is determined as a ratio between the parameter’s mean gradients with respect to a loss function and a corresponding variance”
As drafted, is an additional element that does not apply an exception for the abstract ideas in a meaningful way. See MPEP 2106.05(e).
Therefore, the additional elements do not integrate the abstract ideas into a practical application.
Step 2B Analysis: The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to the integration of the abstract ideas into a practical application, all of the additional elements do not apply the exception in a meaningful way. The claim is not patent eligible.
Claim Rejections - 35 USC § 103
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 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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Claims 1-3, 10-13, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang et al. (Removing the Feature Correlation Effect of Multiplicative Noise) (“Zhang”) in view of Zhao et al. (U.S. Patent Publication No. 2022/0180627) (“Zhao”).
Regarding claim 1, Zhang teaches a computer-implemented method for training a model (Zhang Section 1 Introduction “Our analysis is based on a simple assumption that, in order to reduce the interference of noise, the training process will try to maximize the signal-to-noise ratio (SNR) of representations.”; Section 2.2 The Feature Correlation Effect “From an information perspective, injecting noise into a neural network can be seen as training the model to solve the task with noisy information pathways.” Zhang provides methods for training a neural network.), comprising: determining a dropout mask (Zhang Section 2.1 Multiplicative Noise “The noise mask, ui, can be sampled from various distributions, as exemplified by Bernoulli, Gaussian, and uniform distributions. We take dropout and DropConnect as examples. For dropout, let mi be the dropout mask sampled from a Bernoulli distribution, Bern(p), then the equivalent multiplicative noise is given by ui = mi/p… Thus, we denote the dropout mask by mij, where j ∈ Hl+1, then we have uij = mij/p and (Equation (2)).”; Section 3.1 Non-correlating Multiplicative Noise “Due to the use of batch normalization, Eq. (12) is a constant with respect to each sample of zs j, and thus we have ∂E ˆzs j 2 /∂zs j(m) = 0,∀m ∈ B, where B denotes a set of mini-batch samples, and zs j (m) denotes the value of zs j corresponding to sample m.” Zhang provides determining a dropout mask based on noise, as denoted by m, which includes use of a signal to noise ratio denoted by z, wherein a dropout mask is determined based on the calculated signal to noise ratio, in accordance with Equations 1-2, and 8-12, wherein m corresponds to the dropout mask based on a calculated value z.) based on gradient signal to noise ratio (GSNR) of parameters of a neural network model (Zhang Section 1 Introduction “Our analysis is based on a simple assumption that, in order to reduce the interference of noise, the training process will try to maximize the signal-to-noise ratio (SNR) of representations.”; Section 3.1 Non-Correlating Multiplicative Noise “Eq. (8) implies that if we ignore the gradient of the noise component, zn j , the tendency of increasing the SNR of zj will attempt to increase the variance of the signal component, zs j, instead of increasing the feature correlation of the lower layer. However, we find in practice that such modification to the gradient causes optimization difficulties, preventing the training process from converging… We now consider the SNR of the new pre-activation, ˆzj. The signal and noise components of ˆzj are respectively (Equation (10)). For clarity, we define an identity function, AsConst(·), meaning that the argument of the function is considered as a constant during the backpropagation phase, or in other words, its gradient is set to zero by the function.” Zhang provides using noise to determine a dropout mask m, including a signal to noise ratio including use of a gradient for neural network training which is set to zero, corresponding to the gradient signal to noise ratio of neural network parameters to determine a dropout mask, as shown in Equation (10).); training the neural network model with parameters zeroed-out according to the dropout mask (Zhang Section 2.1 Multiplicative Noise “When used for regularization purpose, multiplicative noise is typically applied at training time, and removed at test time. Consequently, the noise should satisfy E[ui] = 1, such that E[˜xi] = xi… Thus, we denote the dropout mask by mij, where j ∈ Hl+1, then we have uij = mij/p and (Equation (2)).”; Section 3.1 Non-Correlating Multiplicative Noise “Due to the use of batch normalization, Eq. (12) is a constant with respect to each sample of zs j, and thus we have ∂E ˆzs j 2 /∂zs j(m) = 0,∀m ∈ B, where B denotes a set of mini-batch samples, and zs j (m) denotes the value of zs j corresponding to sample m.” Zhang provides training a neural network including use of a dropout mask, which zeros out neural network parameters.);
Zhang fails to explicitly teach iteratively updating the dropout mask and performing the training based on the updated dropout mask.
However, Zhao teaches iteratively updating the dropout mask (Zhao [0070] “The present disclosure further proposes improved model training for object recognition, wherein in the training process, an optimized training data set is obtained by adopting the training data optimization method according to an embodiment of the present disclosure, and the optimized training data set is used for model training, the training process will be iteratively performed until the training end condition is satisfied.”; [0144] “In particular, as an example, with reference to the dropout mask for each training sample, the weighting unit may only weight the loss functions for the training samples used for neural network model training”; [0153] “Then, a dropout mask for each training sample is initialized, for example, the dropout mask for each training sample can be set to 1, so as to indicate that all training samples can be used for model training.”; Zhao provides an iterative training process, wherein a dropout mask is determined for each training sample, thus iteratively updating a dropout mask.) and performing the training based on the updated dropout mask (Zhao [0100] “As an example, a corresponding dropout mask can be set for a training sample according to the result of judgment that whether its fluctuation is less than a specific threshold, so that the dropout mask can be utilized for indicating whether the training sample is suitable for subsequent model training.”; [0153] “Then, a dropout mask for each training sample is initialized, for example, the dropout mask for each training sample can be set to 1, so as to indicate that all training samples can be used for model training.” Zhao provides performing iterative training based on an updated dropout mask for each training sample.).
Zhang and Zhao are both considered to be analogous to the claimed invention because they are in the same field of artificial intelligence and more specifically applied to dropout mask neural network training. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Zhang with the above teachings of Zhao. Doing so would provide an improved training data optimization method (Zhao [0063] “In view of this, the present disclosure proposes an improved training data optimization method, which can optimize the training data by utilizing the fluctuation of training samples in the training database and carry out model training based on the optimized training data, thereby realizing improved model training and improving model recognition ability.”).
Regarding claim 2, Zhang in view of Zhao teaches wherein determining the dropout mask includes determining a dropout ratio that determines a number of parameters to zero-out (Zhang Section 2.1 Multiplicative Noise “Thus, we denote the dropout mask by mij, where j ∈ Hl+1, then we have uij = mij/p and (Equation (2)).”; Section 3.1 Non-Correlating Multiplicative Noise “Concretely, we apply batch normalization to zj as Equation (9)… We now consider the SNR of the new pre-activation, ˆzj. The signal and noise components of ˆzj are respectively (Equation (10)). For clarity, we define an identity function, AsConst(·), meaning that the argument of the function is considered as a constant during the backpropagation phase, or in other words, its gradient is set to zero by the function… Due to the use of batch normalization, Eq. (12) is a constant with respect to each sample of zs j, and thus we have ∂E ˆzs j 2 /∂zs j(m) = 0” Zhang provides determining a dropout and a signal to noise ratio and setting the gradient to zero by the function, corresponding to determining a dropout ratio that determines a number of parameters to zero-out, wherein the ratios of Equation 9 and 10 corresponds to a dropout ratio that determines a number of parameters to zero-out.).
It would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Zhang in view of Zhao for the same reasons disclosed above in the rejection of claim 1.
Regarding claim 3, Zhang in view of Zhao teaches wherein the dropout ratio varies for different parts of the neural network model (Zhang Section 3.1 Non-Correlating Multiplicative Noise “From Eq. (5) we observe that, if we consider the denominator E zn j 2 as a constant, i.e., ignore its gradient during training, the objective function is equivalent to (Equation (8). Eq. (8) implies that if we ignore the gradient of the noise component, zn j , the tendency of increasing the SNR of zj will attempt to increase the variance of the signal component, zs j, instead of increasing the feature correlation of the lower layer. However, we find in practice that such modification to the gradient causes optimization difficulties, preventing the training process from converging… The signal and noise components of ˆzj are respectively: Equation (10).” Zhang provides the signal to noise ratio increasing in value during training for a lower layer, corresponding the dropout ratio varies for different parts of the neural network, wherein the lower layers corresponds to the different part of the neural network and wherein the ratio of Equation 10 corresponds to the dropout ratio.).
It would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Zhang in view of Zhao for the same reasons disclosed above in the rejection of claim 2.
Regarding claim 10, Zhang in view of Zhao teaches wherein the GSNR of a parameter is determined as a ratio between the parameter’s mean gradients with respect to a loss function and a corresponding variance (Zhang Section 3.1 Non-Correlating Multiplicative Noise “From Eq. (5) we observe that, if we consider the denominator E zn j 2 as a constant, i.e., ignore its gradient during training, the objective function is equivalent to Equation (8)…However, we find in practice that such modification to the gradient causes optimization difficulties, preventing the training process from converging. Fortunately, by using batch normalization, the remedy for this problem is surprisingly simple… We neglect the small difference between the true mean/variance, and the sample mean/variance, and adopt the same notation for simplicity. Note that in Eq. (9), zj is normalized using the statistics of zs j, which is slightly different from standard batch normalization. We now consider the SNR of the new pre-activation, ˆzj. The signal and noise components of ˆzj are respectively Equation (10).” Zhang provides the signal to noise ratio of Equation 10, which includes a ratio of the gradients mean and variance (see Equation 10), wherein the objective function of Equation (8) corresponds to the loss function, and wherein the ratio of Equation 10 is the ratio between the parameter’s mean gradients with respect to a loss function and a corresponding variance.).
It would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Zhang in view of Zhao for the same reasons disclosed above in the rejection of claim 1.
Regarding claim 11, it is the system embodiment of claim 1 with similar limitations to claim 1 and is rejected using the same reasoning found above in the rejection of claim 1. Further, Zhao teaches a system for training a model, comprising: a hardware processor (Zhao [0138] “It should be noted that in addition to including a plurality of units, the above-mentioned training apparatus may be implemented in various other forms, such as a general-purpose processor or a dedicated processing circuit such as an ASIC.” Zhao provides a training apparatus including a hardware processor.); and a memory that stores a computer program (Zhao [0138] “In addition, the training apparatus may carry a program (software) for operating a circuit (hardware) or a central processing device. The program can be stored in a storage (such as arranged in the storage) or an external storage medium connected from the outside, and downloaded via a network (such as the Internet).” Zhao provides memory storing a computer program.).
Zhang and Zhao are both considered to be analogous to the claimed invention because they are in the same field of artificial intelligence and more specifically applied to dropout mask neural network training. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Zhang with the above teachings of Zhao. Doing so would provide an improved training data optimization method (Zhao [0063] “In view of this, the present disclosure proposes an improved training data optimization method, which can optimize the training data by utilizing the fluctuation of training samples in the training database and carry out model training based on the optimized training data, thereby realizing improved model training and improving model recognition ability.”).
Regarding claim 12, the rejection of claim 11 is incorporated herein. Further, the limitations in this claim are taught by Zhang in view of Zhao for the same reasons disclosed above in the rejection of claim 2.
Regarding claim 13, the rejection of claim 12 is incorporated herein. Further, the limitations in this claim are taught by Zhang in view of Zhao for the same reasons disclosed above in the rejection of claim 3.
Regarding claim 20, the rejection of claim 11 is incorporated herein. Further, the limitations in this claim are taught by Zhang in view of Zhao for the same reasons disclosed above in the rejection of claim 10.
Claims 4-5 and 14-15 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang et al. (Removing the Feature Correlation Effect of Multiplicative Noise) (“Zhang”) in view of Zhao et al. (U.S. Patent Publication No. 2022/0180627) (“Zhao”) in further view of Tseng et al. (Regularizing Meta-Learning via Gradient Dropout) (“Tseng”).
Regarding claim 4, Zhang in view Zhao teaches the method of claim 1, as discussed above in the rejection of claim 1, but fails to teach wherein determining the dropout mask includes performing meta-training and meta-testing to update a loss function.
However, Tseng teaches wherein determining the dropout mask includes performing meta-training and meta-testing to update a loss function (Tseng Section 3 Gradient Dropout Regularization “Before introducing details of our proposed dropout regularization on gradients, we first review the gradient-based meta-learning framework.”; Section 3.1 Preliminaries for Meta-Learning “In meta-learning, multiple tasks T = {T1,T2,...,Tn} are divided into meta training T train, meta-validation T val, and meta-testing T test sets… Given a novel task and a parametric model fθ, the objective of a gradient based approach during the meta-training stage is to minimize the prediction loss Lq on the query set Dq according to the signals provided from the support set Ds, and thus the model fθ can be adapted.” Tseng provides a dropout regularization method including meta-learning, which includes meta-training and meta-testing to minimize (update) a loss function.).
Zhang, Zhao, and Tseng are all considered to be analogous to the claimed invention because they are in the same field of artificial intelligence and more specifically applied to dropout neural network training. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Zhang in view of Zhao with the above teachings of Tseng. Doing so would alleviate the risk of overfitting for gradient-based meta-learning (Tseng Abstract “In this paper, we introduce a simple yet effective method to alleviate the risk of overfitting for gradient-based meta-learning.”).
Regarding claim 5, Zhang in view of Zhao teaches the method of claim 3, as discussed above in the rejection of claim 3, but fails to explicitly teach wherein determining the dropout mask includes determining a gradient of the loss function.
However, Tseng teaches wherein determining the dropout mask includes determining a gradient of the loss function (Tseng Gradient Dropout Regularization “Before introducing details of our proposed dropout regularization on gradients, we first review the gradient-based meta-learning framework.”; Section 3.1 Preliminaries for Meta-Learning ““In meta-learning, multiple tasks T = {T1,T2,...,Tn} are divided into meta training T train, meta-validation T val, and meta-testing T test sets… Given a novel task and a parametric model fθ, the objective of a gradient based approach during the meta-training stage is to minimize the prediction loss Lq on the query set Dq according to the signals provided from the support set Ds, and thus the model fθ can be adapted.” Tseng provides determining dropout including a gradient based approach to minimize a prediction loss function, thus including a gradient of the loss function.).
Zhang, Zhao, and Tseng are all considered to be analogous to the claimed invention because they are in the same field of artificial intelligence and more specifically applied to dropout neural network training. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Zhang in view of Zhao with the above teachings of Tseng. Doing so would alleviate the risk of overfitting for gradient-based meta-learning (Tseng Abstract “In this paper, we introduce a simple yet effective method to alleviate the risk of overfitting for gradient-based meta-learning.”).
Regarding claim 14, the rejection of claim 11 is incorporated herein. Further, the limitations in this claim are taught by Zhang in view of Zhao in further view of Tseng for the same reasons disclosed above in the rejection of claim 4.
Regarding claim 15, the rejection of claim 14 is incorporated herein. Further, the limitations in this claim are taught by Zhang in view of Zhao in further view of Tseng for the same reasons disclosed above in the rejection of claim 5.
Claims 6 and 16 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang et al. (Removing the Feature Correlation Effect of Multiplicative Noise) (“Zhang”) in view of Zhao et al. (U.S. Patent Publication No. 2022/0180627) (“Zhao”) in further view of Hospedales et al. (Meta-Learning in Neural Networks: A Survey) (“Hospedales”).
Regarding claim 6, Zhang in view of Zhao teaches the of claim 3, as discussed above in the rejection of claim 3, but fails to teach wherein performing meta-training and meta-testing includes selecting meta-training batch subset and a meta-testing batch subset, with examples of the meta-testing batch subset being selected according to their distance from the meta-training batch subset.
However, Hospedales teaches wherein performing meta-training and meta-testing includes selecting meta-training batch subset and a meta-testing batch subset (Hospedales Section 2.1 Formalized Meta-Learning “Formally, we denote the set of M source tasks used in the meta training stage as Dsource = Dtrain source, Dval source, where each task has both training and validation data. Often, the source train and validation datasets are respectively called support and query sets. The meta-training step of ‘learning how to learn’ can be written as: Equation (3)… In the meta-testing stage we use the learned meta-knowledge v to train the base model on each previously unseen target task i: Equation (4).” Hospedales provides a meta-learning approach including meta-training and meta-testing, wherein the meta-training subset is selected in accordance with Equation (3), and the meta-testing subset is selected in accordance with Equation (4).), with examples of the meta-testing batch subset being selected according to their distance from the meta-training batch subset (Hospedales Section 2.1 Formalized Meta-Learning “Now we denote the set of Q target tasks used in the meta-testing stage as Dtarget = Dtrain target, Dtest target, where each task has both training and test data. In the meta-testing stage we use the learned meta-knowledge v to train the base model on each previously unseen target task i: Equation (4).” Hospedales provides selecting a meta-testing set in accordance with Equation (4), wherein Equation(4) uses an argmin function of the meta-training set defined by Equation (3), which produces a smallest/shortest distance from the meta-training set, thus selecting the meta-testing set based on a distance to the meta-training set.).
Zhang, Zhao, and Hospedales are all considered to be analogous to the claimed invention because they are in the same field of artificial intelligence and more specifically applied to neural network training. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Zhang in view of Zhao with the above teachings of Hospedales. Doing so would allow for learning a general purpose learning algorithm that generalizes across tasks, and enables each new task to be learned better than the last (Hospedales Section 2.1 Formalized Meta-Learning “Meta-Learning: Task-Distribution View. A common view of meta-learning is to learn a general purpose learning algorithm that generalizes across tasks, and ideally enables each new task to be learned better than the last.”).
Regarding claim 16, the rejection of claim 14 is incorporated herein. Further, the limitations in this claim are taught by Zhang in view of Zhao in further view of Hospedales for the same reasons disclosed above in the rejection of claim 6.
Claims 7 and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang et al. (Removing the Feature Correlation Effect of Multiplicative Noise) (“Zhang”) in view of Zhao et al. (U.S. Patent Publication No. 2022/0180627) (“Zhao”) in further view of Lee et al. (Drop to Adapt: Learning Discriminative Features for Unsupervised Domain Adaptation) (“Lee”).
Regarding claim 7, Zhang in view of Zhao teaches the method of claim 1, as discussed above in the rejection of claim 1, but fails to teach wherein training the neural network model includes a training dataset of examples in a first domain and wherein the dropout mask causes the training to better accommodate testing examples from second domains that are not included in the training dataset.
However, Lee teaches wherein training the neural network model includes a training dataset of examples in a first domain (Lee Section 1 Introduction “The advent of deep neural networks (DNNs) has shown exceptional performances on various visual recognition tasks using large-scale datasets [8, 21, 13]. Training a DNN model begins with curating data and its associated label. To take full advantage of synthetic datasets, domain adaptation has become an active research area. In the domain adaptation setting, we leverage rich annotations on a source domain to achieve strong performance on a target do main regardless of poor annotations. Nevertheless, a model trained only on the source domain provides disappointing outcomes when the target domain shows inherently different characteristics. This issue is known as domain shift and is one of the main reasons for performance drops on the target domain.” Lee provides training a deep neural network on a source domain, which is different from a target domain and includes a training dataset in a first domain (i.e., the source domain).) and wherein the dropout mask causes the training to better accommodate testing examples from second domains that are not included in the training dataset (Lee Section 3.1 Unsupervised Domain Adaptation “We employ a feature extractor f(x;mf), where mf represents a dropout mask which can be applied at an arbitrary layer of the feature extractor.”; Section 6 Conclusion “We presented a simple yet effective method for unsupervised domain adaptation despite large domain shifts. With two types of proposed adversarial dropout modules, EAdD and CAdD, we enforced the cluster assumption on the target domain. The proposed methods are easily integrated into existing deep learning architectures. Through extensive experiments on various small and large datasets, we demonstrated the effectiveness of the proposed method on two domain adaptation tasks, and in all cases we achieved significant improvement as compared to the source-only model and the state-of-the-art results” Lee provides using adversarial dropout modules for domain adaption, which includes the use of a dropout mask to improve domain adaptation tasks, which include a source domain and a target domain, which respectively correspond to a first and second domain, and wherein the significant improvement as compared to the source-only model in the domain adaption task corresponds to the dropout mask causing the training to better accommodate testing examples from a second domain (i.e., the target domain) that are not included in the training dataset (i.e., the source domain).).
Zhang, Zhao, and Lee are all considered to be analogous to the claimed invention because they are in the same field of artificial intelligence and more specifically applied to neural network dropout training. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Zhang in view of Zhao with the above teachings of Lee. Doing so would achieve significant improvements as compared to the source-only model and the state-of-the-art results (Lee Section 6 Conclusion “Through extensive experiments on various small and large datasets, we demonstrated the effectiveness of the proposed method on two domain adaptation tasks, and in all cases we achieved significant improvement as compared to the source-only model and the state-of-the-art results”).
Regarding claim 17, the rejection of claim 11 is incorporated herein. Further, the limitations in this claim are taught by Zhang in view of Zhao in further view of Lee for the same reasons disclosed above in the rejection of claim 7.
Claims 8 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang et al. (Removing the Feature Correlation Effect of Multiplicative Noise) (“Zhang”) in view of Zhao et al. (U.S. Patent Publication No. 2022/0180627) (“Zhao”) in further view of Lee et al. (Drop to Adapt: Learning Discriminative Features for Unsupervised Domain Adaptation) (“Lee”) in further view of Wulfmeier et al. (Incremental Adversarial Domain Adaptation for Continually Changing Environments) (“Wulfmeier”).
Regarding claim 8, Zhang in view of Zhao in further view of Lee teaches the method of claim 7 as discussed above in the rejection of claim 7, but fails to explicitly teach wherein the training dataset includes images and wherein the first domain and the second domains differ according to environmental conditions or geographic location.
However, Wulfmeier teaches wherein the training dataset includes images (Wulfmeier Section III Methods “The parameters of both of these modules remain unchanged during training for domain adaptation, which enables us to keep source performance unaffected (an approach suggested for regular ADA in [8]). Only the target encoder Et and discriminator D are trained via their respective objectives LEt and LD in Equations 1 and 2 to align the target and source encoder feature spaces. Let fs = Es(is,θEs ) and ft = Et(it,θEt ) respectively denote the feature encoding of source and target images is and it.” Wulfmeier provides training including source and target images, wherein the source images correspond to a training dataset including images.) and wherein the first domain and the second domains differ according to environmental conditions or geographic location (Wulfmeier Section III Methods “Incremental Adversarial Domain Adaptation addresses the problem of degraded model performance due to continuously shifting environmental conditions. This includes changes caused by weather and lighting occurring in outdoor scenarios. Compared to the regular single-step domain adaptation paradigm, we benefit in applications building on continual deployment through exploitation of the incremental changes that integrate to large domain shifts.” Wulfmeier provides domain shift including continuously shifting environmental conditions, corresponding to differing environmental conditions between a first and second domain.).
Zhang, Zhao, Lee, and Wulfmeier are all considered to be analogous to the claimed invention because they are in the same field of artificial intelligence and more specifically applied to neural network training. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Zhang in view of Zhao in further view of Lee with the above teachings of Wulfmeier. Doing so would provide benefits as models can be adapted to the currently encountered environment and learn from data unavailable during offline training (Wulfmeier Section VI Conclusion “The field of continual training during deployment provides many possible benefits as models can be adapted to the currently encountered environment and learn from data unavailable during offline training.”).
Regarding claim 18, the rejection of claim 17 is incorporated herein. Further, the limitations in this claim are taught by Zhang in view of Zhao in further view of Lee and Wulfmeier for the same reasons disclosed above in the rejection of claim 8.
Claims 9 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Zhang et al. (Removing the Feature Correlation Effect of Multiplicative Noise) (“Zhang”) in view of Zhao et al. (U.S. Patent Publication No. 2022/0180627) (“Zhao”) in further view of Sainath et al. (U.S. Patent Publication No. 2015/0161993) (“Sainath”).
Regarding claim 9, Zhang in view of Zhao teaches the method of claim 1, as discussed above in the rejection of claim 1, but fails to teach wherein training the neural network model with parameters zeroed-out includes performing a feed-forward operation where parameters designated by the dropout mask are omitted.
However, Sainath teaches wherein training the neural network model with parameters zeroed-out (Sainath [0057] “Dropout is a technique to prevent over-fitting during neural network training.”; [0061] “It was experimentally confirmed that using a dropout probability of p=0.5 in the 3rd and 4th layers is reasonable, and the dropout in all other layers is zero.” Sainath provides neural network training including dropout, which includes zeroing out.) includes performing a feed-forward operation where parameters designated by the dropout mask are omitted (Sainath [0057] “Dropout is a technique to prevent over-fitting during neural network training. Specifically, during a feed-forward operation in neural network training, dropout omits each hidden unit randomly with probability p. This prevents complex co-adaptations between hidden units, forcing hidden units to not depend on other units. Specifically, using dropout the activation y.sup.l at layer l is given by Equation 6” Sainath provides a feed-forward operation including dropout, where the dropout omits each hidden unit randomly with probability p, corresponding to the parameters designated by the dropout mask that are omitted.).
Zhang, Zhao, and Sainath are all considered to be analogous to the claimed invention because they are in the same field of artificial intelligence and more specifically applied to neural network dropout training. Therefore, it would have been obvious to someone of ordinary skill in the art before the effective filing date of the claimed invention to have modified Zhang in view of Zhao with the above teachings of Sainath. Doing so would prevent complex co-adaptations between hidden units, forcing hidden units to not depend on other units (Sainath [0057] “Specifically, during a feed-forward operation in neural network training, dropout omits each hidden unit randomly with probability p. This prevents complex co-adaptations between hidden units, forcing hidden units to not depend on other units.”).
Regarding claim 19, the rejection of claim 11 is incorporated herein. Further, the limitations in this claim are taught by Zhang in view of Zhao in further view of Sainath for the same reasons disclosed above in the rejection of claim 9.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to KURT NICHOLAS PRESSLY whose telephone number is (703)756-4639. The examiner can normally be reached M-F 8-4.
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, Kamran Afshar can be reached at (571) 272-7796. 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.
/KURT NICHOLAS PRESSLY/Examiner, Art Unit 2125
/KAMRAN AFSHAR/Supervisory Patent Examiner, Art Unit 2125
Prosecution Insights