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 .
DETAILED ACTION
Claims 1-10 are presented for examination in this application, 18/267,730, originally filed 06/15/2023, having an effective filing date of 12/24/2020 via PCT/JP2020/048472. Claims 1, 5, 6, 9, and 10 have been amended. The arguments made against the 35 U.S.C 101 rejections have been considered and were found to be persuasive. The arguments made against the 35 U.S.C 103 rejections have been considered but were not found to be persuasive.
The Examiner cites particular sections in the references as applied to the claims below for the convenience of the applicant(s). Although the specified citations are
representative of the teachings in the art and are applied to the specific limitations within
the individual claim, other passages and figures may apply as well. It is respectfully
requested that, in preparing responses, the applicant(s) fully consider the references in
their entirety as potentially teaching all or part of the claimed, as well as the context of
the passage as taught by the prior art or disclosed by the Examiner.
Information Disclosure Statement
Acknowledgement is made of the information disclosure statements filed 6/10/2026. All patents and non-patent literature have been considered.
Response to Arguments
The arguments made against the 35 U.S.C 103 rejections have been considered but were not found to be persuasive. The 35 U.S.C rejections have been maintained.
35 U.S.C 103
Applicant asserts:
Applicant asserts “independent claims have been amended to recite features that are not taught or suggested by the cited references. Applicant respectfully requests that the rejections under 35 USC 103 be withdrawn.”.
Applicant further asserts “claim 4 is patentable over the cited references on an independent ground. The Office Action cites Hunter for teaching or suggesting" the likelihood ratios of the Nx(N- 1) patterns" as recited in claim 4. However, Hunter describes patterns of pairwise comparisons between objects to be compared. Specifically, the patterns in Hunter concern combinations of comparison objects in a generalized Bradley-Terry model. In contrast, the patterns recited in the5
present application are likelihood-ratio patterns between candidate classes for classification of the same input sequence data. More specifically, the amended claims recite training using a loss function that takes into account, among the N x (N-1) likelihood-ratio patterns between the candidate classes, likelihood ratios in which the correct class is in the numerator. Accordingly, Hunter does not teach or suggest selecting, from among likelihood-ratio patterns between candidate classes for the same input sequence data, likelihood ratios in which the correct class is in the numerator and incorporating the selected likelihood ratios into a loss function for training a machine learning model used for classifying the input sequence data.”
Examiner’s response:
Examiner finds that the primary reference, Yamamoto, covers most of the newly amended limitations found in the independent claims. Yamamoto teaches adjusting parameters, calculating ratios of likelihoods based on classes, classifying the data into a class from two classes and performing learning based on the aforementioned calculations. Yamamoto does not teach series data. Kato, however, does teach likelihood ratios using series data.
In regard to the argument made that Hunter does not teach or suggest selecting, from among likelihood-ratio patterns between candidate classes for the same input sequence data, likelihood ratios in which the correct class is in the numerator and incorporating the selected likelihood ratios into a loss function for training a machine learning model used for classifying the input sequence data, Hunter is not being used to teach this matter. Hunter is being used teach the N(N-1) patterns within the field of calculating likelihood ratios. It has been deemed obvious for a person having ordinary skill in the art using broadest reasonable interpretation to combine Yamamoto, Kato, and Hunter to produce the matter found in claim 4.
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.
Claims 1, 5, 6, 9, and 10 are rejected under 35 U.S.C 103 as being unpatentable over Yamamoto et al. (US20070088548A1 hereinafter, Yamamoto) in view of Kato (US8612227B2 hereinafter, Kato).
Regarding claim 1:
Yamamoto teaches an information processing system comprising: at least one memory that is configured to store instructions; and at least one processor that is configured to execute the instructions to obtain a first plurality of elements included in (see para [0062]: “a central processing unit (CPU) 52 that controls each section of the speech-section detecting device 10 according to a program stored in ROM 52; a random access memory (RAM) 53 that stores therein various data necessary for a control of the speech-section detecting device 10; ”);
calculate a likelihood ratio indicating a likelihood of a class to which the (see para [0039]: “Next, the model comparing unit 108 calculates an evaluation value LR indicative of the likelihood of speech (log-likelihood ratio) using the m-dimensional feature vector and speech/non-speech Gaussian Mixture Model (GMM) acquired through learning in advance (step S108) as follows:LR=g(y|speech)−g(y|nonspeech) (6)where g(|speech) is the log-likelihood of the speech GMM, and g(|nonspeech) is the log-likelihood of the non-speech GMM.”. Also see para [0054]: “Data is classified into either one of the two classes: speech (C1) and non-speech (C2)”);
classify the (see para [0054]: “Data is classified into either one of the two classes: speech (C1) and non-speech (C2)”. Also see para [0060]: “Moreover, the EM algorithm is based on the maximum likelihood criteria of a sample acquired through learning. These methods are not the best to acquire parameters through learning for the speech/non-speech determination.”);
perform learning related to calculation of the likelihood ratio to provide a learned model, by using a loss function in which the likelihood ratio increases when a correct answer class to which the (see para [0056]: “Dk(y:Λ) in Equation 9 is a log-likelihood between gk and gi. Dk(y:Λ) becomes negative when an acoustic signal, which is a sample acquired through learning, is classified as belonging to the right-answer category. On the other hand, Dk(y:Λ) becomes positive when an acoustic signal, which is a sample acquired through learning, is classified as belonging to the wrong-answer category.”. Also see para [0057]: “The loss lk provided by the loss function is closer to 1 (one) when the rate of wrong recognition is larger, and to 0 (zero) when the error rate is smaller. Learning of the parameter set Λ is performed so as to lower the value provided by the loss function”).;
adjust parameters of the learned model to reduce the loss function (see para [0059]: “As explained above, parameters of the transformation matrix P and the speech/non-speech GMM used when an n-dimensional feature vector extracted from the frames is transformed into an m-dimensional vector (m<n) can be adjusted so as to minimize a rate of wrong recognition using the discriminative learning method. Therefore, performance of the speech/non-speech determination can be improved. Furthermore, a speech section can be detected more accurately.”);
obtain a second plurality of elements included in the (see para [0012]: “a second storage unit that stores therein a first parameter of a speech model and a second parameter of a non-speech model, wherein the first parameter and the second parameter are calculated based on the speech/non-speech likelihood; an acquiring unit that acquires an acoustic signal; a dividing unit that divides the acoustic signal into a plurality of frames; an extracting unit that extracts a feature vector from acoustic signals of the frames”);
calculate a likelihood ratio indicating a likelihood of a class to which the (see para [0012]: “wherein the transformation matrix is calculated based on an actual speech/non-speech likelihood calculated from a known sample acquired through learning”)
classify the (see para [0054]: “Data is classified into either one of the two classes: speech (C1) and non-speech (C2)”);
perform learning related to calculation of the likelihood ratio to provide an updated learned model, by using the loss function (see para [0059]: “As explained above, parameters of the transformation matrix P and the speech/non-speech GMM used when an n-dimensional feature vector extracted from the frames is transformed into an m-dimensional vector (m<n) can be adjusted so as to minimize a rate of wrong recognition using the discriminative learning method. Therefore, performance of the speech/non-speech determination can be improved. Furthermore, a speech section can be detected more accurately.”).
Yamamoto does not explicitly teach series data with likelihood ratios.
Kato, however, analogously teaches series data with likelihood ratios (see col 7 lines 32-38: “A likelihood calculation unit 153 calculates an acoustic likelihood by matching time series data of acoustic feature parameters against a lexical tree stored in a second database 20 and an acoustic model stored in a third database 21 in the self-transition and LR transition to determine an accumulated likelihood by accumulating the acoustic likelihood in a time direction..”)
Before the effective filing date of the claimed invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Yamamoto and Kato before him or her, to modify the system of claim 1 to include attributes of series data with likelihood ratios in order to operate on ordered data that has regular intervals (see col 7 lines 19-22: “ Acoustic feature parameters are a feature vector obtained by analyzing an input speech at regular intervals (for example, 10 ms; hereinafter, denoted as frames). Therefore, the audio signal is converted into a time series”).
Regarding claim 5:
Yamamoto in view of Kato teaches the system of claim 1.
Yamamoto further teaches wherein the loss function includes a sigmoid function as a nonlinear function that affects the likelihood ratio (see para [0056]: “ loss lk due to a classification error (y;Λ) is defined by Equation 10. Also see equation 10:
PNG
media_image1.png
49
506
media_image1.png
Greyscale
).
Regarding claim 6:
Yamamoto in view of Kato teaches the system of claim 1.
Yamamoto further teaches wherein the loss function includes logistic function as a nonlinear function that affects the likelihood ratio (see para [0056]: “ loss lk due to a classification error (y;Λ) is defined by Equation 10. Also see equation 10:
PNG
media_image1.png
49
506
media_image1.png
Greyscale
).
Regarding claim 9:
Yamamoto teaches an information processing system method to obtain a first plurality of elements included in (see para [0062]: “a central processing unit (CPU) 52 that controls each section of the speech-section detecting device 10 according to a program stored in ROM 52; a random access memory (RAM) 53 that stores therein various data necessary for a control of the speech-section detecting device 10; ”);
calculating a likelihood ratio indicating a likelihood of a class to which the (see para [0039]: “Next, the model comparing unit 108 calculates an evaluation value LR indicative of the likelihood of speech (log-likelihood ratio) using the m-dimensional feature vector and speech/non-speech Gaussian Mixture Model (GMM) acquired through learning in advance (step S108) as follows:LR=g(y|speech)−g(y|nonspeech) (6)where g(|speech) is the log-likelihood of the speech GMM, and g(|nonspeech) is the log-likelihood of the non-speech GMM.”. Also see para [0054]: “Data is classified into either one of the two classes: speech (C1) and non-speech (C2)”);
classifying the (see para [0054]: “Data is classified into either one of the two classes: speech (C1) and non-speech (C2)”. Also see para [0060]: “Moreover, the EM algorithm is based on the maximum likelihood criteria of a sample acquired through learning. These methods are not the best to acquire parameters through learning for the speech/non-speech determination.”);
performing learning related to calculation of the likelihood ratio to provide a learned model, by using a loss function in which the likelihood ratio increases when a correct answer class to which the (see para [0056]: “Dk(y:Λ) in Equation 9 is a log-likelihood between gk and gi. Dk(y:Λ) becomes negative when an acoustic signal, which is a sample acquired through learning, is classified as belonging to the right-answer category. On the other hand, Dk(y:Λ) becomes positive when an acoustic signal, which is a sample acquired through learning, is classified as belonging to the wrong-answer category.”. Also see para [0057]: “The loss lk provided by the loss function is closer to 1 (one) when the rate of wrong recognition is larger, and to 0 (zero) when the error rate is smaller. Learning of the parameter set Λ is performed so as to lower the value provided by the loss function”).;
adjusting parameters of the learned model to reduce the loss function (see para [0059]: “As explained above, parameters of the transformation matrix P and the speech/non-speech GMM used when an n-dimensional feature vector extracted from the frames is transformed into an m-dimensional vector (m<n) can be adjusted so as to minimize a rate of wrong recognition using the discriminative learning method. Therefore, performance of the speech/non-speech determination can be improved. Furthermore, a speech section can be detected more accurately.”);
obtain a second plurality of elements included in the (see para [0012]: “a second storage unit that stores therein a first parameter of a speech model and a second parameter of a non-speech model, wherein the first parameter and the second parameter are calculated based on the speech/non-speech likelihood; an acquiring unit that acquires an acoustic signal; a dividing unit that divides the acoustic signal into a plurality of frames; an extracting unit that extracts a feature vector from acoustic signals of the frames”);
calculating a likelihood ratio indicating a likelihood of a class to which the (see para [0012]: “wherein the transformation matrix is calculated based on an actual speech/non-speech likelihood calculated from a known sample acquired through learning”)
classifying the (see para [0054]: “Data is classified into either one of the two classes: speech (C1) and non-speech (C2)”);
performing learning related to calculation of the likelihood ratio to provide an updated learned model, by using the loss function (see para [0059]: “As explained above, parameters of the transformation matrix P and the speech/non-speech GMM used when an n-dimensional feature vector extracted from the frames is transformed into an m-dimensional vector (m<n) can be adjusted so as to minimize a rate of wrong recognition using the discriminative learning method. Therefore, performance of the speech/non-speech determination can be improved. Furthermore, a speech section can be detected more accurately.”).
Yamamoto does not explicitly teach series data with likelihood ratios.
Kato, however, analogously teaches series data with likelihood ratios (see col 7 lines 32-38: “A likelihood calculation unit 153 calculates an acoustic likelihood by matching time series data of acoustic feature parameters against a lexical tree stored in a second database 20 and an acoustic model stored in a third database 21 in the self-transition and LR transition to determine an accumulated likelihood by accumulating the acoustic likelihood in a time direction..”)
Before the effective filing date of the claimed invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Yamamoto and Kato before him or her, to modify the method of claim 9 to include attributes of series data with likelihood ratios in order to operate on ordered data that has regular intervals (see col 7 lines 19-22: “ Acoustic feature parameters are a feature vector obtained by analyzing an input speech at regular intervals (for example, 10 ms; hereinafter, denoted as frames). Therefore, the audio signal is converted into a time series”).
Regarding claim 10:
Yamamoto teaches an information processing method including: obtaining a first plurality of elements included in (see para [0062]: “a central processing unit (CPU) 52 that controls each section of the speech-section detecting device 10 according to a program stored in ROM 52; a random access memory (RAM) 53 that stores therein various data necessary for a control of the speech-section detecting device 10; ”);
calculating a likelihood ratio indicating a likelihood of a class to which the (see para [0039]: “Next, the model comparing unit 108 calculates an evaluation value LR indicative of the likelihood of speech (log-likelihood ratio) using the m-dimensional feature vector and speech/non-speech Gaussian Mixture Model (GMM) acquired through learning in advance (step S108) as follows:LR=g(y|speech)−g(y|nonspeech) (6)where g(|speech) is the log-likelihood of the speech GMM, and g(|nonspeech) is the log-likelihood of the non-speech GMM.”. Also see para [0054]: “Data is classified into either one of the two classes: speech (C1) and non-speech (C2)”);
classifying the (see para [0054]: “Data is classified into either one of the two classes: speech (C1) and non-speech (C2)”. Also see para [0060]: “Moreover, the EM algorithm is based on the maximum likelihood criteria of a sample acquired through learning. These methods are not the best to acquire parameters through learning for the speech/non-speech determination.”);
performing learning related to calculation of the likelihood ratio to provide a learned model, by using a loss function in which the likelihood ratio increases when a correct answer class to which the (see para [0056]: “Dk(y:Λ) in Equation 9 is a log-likelihood between gk and gi. Dk(y:Λ) becomes negative when an acoustic signal, which is a sample acquired through learning, is classified as belonging to the right-answer category. On the other hand, Dk(y:Λ) becomes positive when an acoustic signal, which is a sample acquired through learning, is classified as belonging to the wrong-answer category.”. Also see para [0057]: “The loss lk provided by the loss function is closer to 1 (one) when the rate of wrong recognition is larger, and to 0 (zero) when the error rate is smaller. Learning of the parameter set Λ is performed so as to lower the value provided by the loss function”).
adjusting parameters of the learned model to reduce the loss function (see para [0059]: “As explained above, parameters of the transformation matrix P and the speech/non-speech GMM used when an n-dimensional feature vector extracted from the frames is transformed into an m-dimensional vector (m<n) can be adjusted so as to minimize a rate of wrong recognition using the discriminative learning method. Therefore, performance of the speech/non-speech determination can be improved. Furthermore, a speech section can be detected more accurately.”);
obtain a second plurality of elements included in the (see para [0012]: “a second storage unit that stores therein a first parameter of a speech model and a second parameter of a non-speech model, wherein the first parameter and the second parameter are calculated based on the speech/non-speech likelihood; an acquiring unit that acquires an acoustic signal; a dividing unit that divides the acoustic signal into a plurality of frames; an extracting unit that extracts a feature vector from acoustic signals of the frames”);
calculating a likelihood ratio indicating a likelihood of a class to which the (see para [0012]: “wherein the transformation matrix is calculated based on an actual speech/non-speech likelihood calculated from a known sample acquired through learning”)
classifying the (see para [0054]: “Data is classified into either one of the two classes: speech (C1) and non-speech (C2)”);
performing learning related to calculation of the likelihood ratio to provide a learned model, by using the loss function (see para [0059]: “As explained above, parameters of the transformation matrix P and the speech/non-speech GMM used when an n-dimensional feature vector extracted from the frames is transformed into an m-dimensional vector (m<n) can be adjusted so as to minimize a rate of wrong recognition using the discriminative learning method. Therefore, performance of the speech/non-speech determination can be improved. Furthermore, a speech section can be detected more accurately.”).
adjusting parameters of the learned model to reduce the loss function (see para [0059]: “As explained above, parameters of the transformation matrix P and the speech/non-speech GMM used when an n-dimensional feature vector extracted from the frames is transformed into an m-dimensional vector (m<n) can be adjusted so as to minimize a rate of wrong recognition using the discriminative learning method. Therefore, performance of the speech/non-speech determination can be improved. Furthermore, a speech section can be detected more accurately.”);
obtain a second plurality of elements included in the series data (see para [0012]: “a second storage unit that stores therein a first parameter of a speech model and a second parameter of a non-speech model, wherein the first parameter and the second parameter are calculated based on the speech/non-speech likelihood; an acquiring unit that acquires an acoustic signal; a dividing unit that divides the acoustic signal into a plurality of frames; an extracting unit that extracts a feature vector from acoustic signals of the frames”);
calculating a likelihood ratio indicating a likelihood of a class to which the (see para [0012]: “wherein the transformation matrix is calculated based on an actual speech/non-speech likelihood calculated from a known sample acquired through learning”)
classifying the (see para [0054]: “Data is classified into either one of the two classes: speech (C1) and non-speech (C2)”);
performing learning related to calculation of the likelihood ratio to provide an updated learned model, by using the loss function (see para [0059]: “As explained above, parameters of the transformation matrix P and the speech/non-speech GMM used when an n-dimensional feature vector extracted from the frames is transformed into an m-dimensional vector (m<n) can be adjusted so as to minimize a rate of wrong recognition using the discriminative learning method. Therefore, performance of the speech/non-speech determination can be improved. Furthermore, a speech section can be detected more accurately.”).
Yamamoto does not explicitly teach series data with likelihood ratios or a non-transitory recording medium.
Kato, however, analogously teaches series data with likelihood ratios (see col 7 lines 32-38: “A likelihood calculation unit 153 calculates an acoustic likelihood by matching time series data of acoustic feature parameters against a lexical tree stored in a second database 20 and an acoustic model stored in a third database 21 in the self-transition and LR transition to determine an accumulated likelihood by accumulating the acoustic likelihood in a time direction..”) and
a non-transitory recording medium (see claim 8: “A non-transitory computer-readable recording medium storing a pattern recognition program, executed by a computer to perform the method of claim 7.”)
Before the effective filing date of the claimed invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Yamamoto and Kato before him or her, to modify the non-transitory recording medium of claim 10 to include attributes of series data with likelihood ratios in order to operate on ordered data that has regular intervals (see col 7 lines 19-22: “ Acoustic feature parameters are a feature vector obtained by analyzing an input speech at regular intervals (for example, 10 ms; hereinafter, denoted as frames). Therefore, the audio signal is converted into a time series”).
Claims 2, 3, and 4 are rejected under 35 U.S.C 103 as being unpatentable over Yamamoto et al. (US20070088548A1 hereinafter, Yamamoto) in view of Kato (US8612227B2 hereinafter, Kato) in further view of Hunter (“MM Algorithms for Generalized Bradley-Terry Models” hereinafter, Hunter).
Regarding claim 2:
Yamamoto in view of Kato teaches the system of claim 1.
Yamamoto further teaches wherein the at least one processor is configured to execute the instructions to perform the learning by using a loss function that takes into account the likelihood ratios of (see para [0039]: “Next, the model comparing unit 108 calculates an evaluation value LR indicative of the likelihood of speech (log-likelihood ratio) using the m-dimensional feature vector and speech/non-speech Gaussian Mixture Model (GMM) acquired through learning in advance (step S108) as follows:LR=g(y|speech)−g(y|nonspeech) (6)where g(|speech) is the log-likelihood of the speech GMM, and g(|nonspeech) is the log-likelihood of the non-speech GMM.”. Also see para [0057]: “The loss lk provided by the loss function is closer to 1 (one) when the rate of wrong recognition is larger, and to 0 (zero) when the error rate is smaller. Learning of the parameter set Λ is performed so as to lower the value provided by the loss function”).
Yamamoto does not explicitly teach likelihood ratios of Nx(N-1) patterns.
Hunter, however, analogously teaches likelihood ratios of Nx(N-1) patterns (see pg. 390 eq. 15. Also see pg. 388: “One feature of the function Qk(γ) defined in (10) that makes it easier to maximize than the original log-likelihood is the fact that it separates the components of the parameter vector γ.”).
Before the effective filing date of the claimed invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Yamamoto, Kato, and Hunter before him or her, to modify the system of claim 2 to include attributes of likelihood ratios of Nx(N-1) patterns in order to aid in maximizing the log-likelihood ratios (see pg. 388-389 section 3: “One feature of the function Qk(γ) defined in (10) that makes it easier to maximize than the original log-likelihood is the fact that it separates the components of the parameter vector γ.”).
Regarding claim 3:
Yamamoto in view of Kato in further view of Hunter teaches the system of claim 1.
Yamamoto does not explicitly teach wherein the at least one processor is configured to execute the instructions to perform the learning by using a loss function that takes into account a part of the likelihood ratios of the Nx(N-1) patterns.
Hunter, however, analogously teaches wherein the at least one processor is configured to execute the instructions to perform the learning by using a loss function that takes into account a part of the likelihood ratios of the Nx(N-1) patterns (see pg. 390 eq. 15. Also see pg. 388: “One feature of the function Qk(γ) defined in (10) that makes it easier to maximize than the original log-likelihood is the fact that it separates the components of the parameter vector γ.”).
Before the effective filing date of the claimed invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Yamamoto, Kato, and Hunter before him or her, to modify the system of claim 3 to include attributes of performing the learning by using a loss function that takes into account a part of the likelihood ratios of the Nx(N-1) patterns in order to aid in maximizing the log-likelihood ratios (see pg. 388-389 section 3: “One feature of the function Qk(γ) defined in (10) that makes it easier to maximize than the original log-likelihood is the fact that it separates the components of the parameter vector γ.”).
Regarding claim 4:
Yamamoto in view of Kato in further view of Hunter teaches the system of claim 3.
Yamamoto further teaches wherein the at least one processor is configured to execute the instructions to perform the learning by using a loss function that takes into account the likelihood ratio in which the correct answer class is in the numerator, out of the (see para [0039]: “Next, the model comparing unit 108 calculates an evaluation value LR indicative of the likelihood of speech (log-likelihood ratio) using the m-dimensional feature vector and speech/non-speech Gaussian Mixture Model (GMM) acquired through learning in advance (step S108) as follows:LR=g(y|speech)−g(y|nonspeech) (6)where g(|speech) is the log-likelihood of the speech GMM, and g(|nonspeech) is the log-likelihood of the non-speech GMM.”).
Yamamoto does not explicitly teach likelihood ratios of Nx(N-1) patterns.
Hunter, however, analogously teaches likelihood ratios of Nx(N-1) patterns (see pg. 390 eq. 15. Also see pg. 388: “One feature of the function Qk(γ) defined in (10) that makes it easier to maximize than the original log-likelihood is the fact that it separates the components of the parameter vector γ.”).
Before the effective filing date of the claimed invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Yamamoto, Kato, and Hunter before him or her, to modify the system of claim 4 to include attributes of likelihood ratios of Nx(N-1) patterns in order to aid in maximizing the log-likelihood ratios (see pg. 388-389 section 3: “One feature of the function Qk(γ) defined in (10) that makes it easier to maximize than the original log-likelihood is the fact that it separates the components of the parameter vector γ.”).
Claims 7 and 8 is rejected under 35 U.S.C 103 as being unpatentable over Yamamoto et al. (US20070088548A1 hereinafter, Yamamoto) in view of Kato (US8612227B2 hereinafter, Kato) in further view of Hunter (“MM Algorithms for Generalized Bradley-Terry Models” hereinafter, Hunter) and further in view of Varin et al. (“Pairwise likelihood inference for ordinal categorical time series” hereinafter, Varin).
Regarding claim 7:
Yamamoto in view of Kato teaches the system of claim 1.
Yamamoto does not explicitly teach wherein the likelihood ratio is an integrated likelihood ratio that is calculated by taking into account a plurality of individual likelihood ratios that are calculated on the basis of two consecutive elements included in the series data.
Varin, however, analogously teaches wherein the likelihood ratio is an integrated likelihood ratio that is calculated by taking into account a plurality of individual likelihood ratios that are calculated on the basis of two consecutive elements included in the series data (see pg. 2368 section 3: “When we consider an AOP(1) model, in order to compute the first-order pairwise likelihood, we only require the calculation of the following bivariate joint probabilities
PNG
media_image2.png
44
728
media_image2.png
Greyscale
. Whenever , follows a bivariate normal distribution with mean vector
PNG
media_image3.png
46
349
media_image3.png
Greyscale
variances σ^2/(1-ɣ^2)and correlation . Computing the first-order pairwise likelihood requires the approximation (n – 1) of bivariate Gaussian integrals instead of an often prohibitive single n-dimensional Gaussian integral, as in ordinary likelihood inference.”).
Before the effective filing date of the claimed invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Yamamoto, Kato, Hunter, and Varin before him or her, to modify the system of claim 7 to include attributes of wherein the likelihood ratio is an integrated likelihood ratio that is calculated by taking into account a plurality of individual likelihood ratios that are calculated on the basis of two consecutive elements included in the series data in order to not be as prohibitive as ordinary likelihood inferences (see pg. 2368 section 3 : “ Computing the first-order pairwise likelihood requires the approximation of bivariate Gaussian integrals instead of an often prohibitive single n-dimensional Gaussian integral, as in ordinary likelihood inference.”).
Regarding claim 8:
Yamamoto in view of Kato in further view of Hunter and further in view of Varin teaches the system of claim 1.
Yamamoto further teaches wherein the at least one processor is configured to execute the instructions to sequentially obtain a plurality of elements included in the (see para [0033]: “Namely, information that is more effective for performing the speech/non-speech determination is included in the time-varying information as compared to information included in the feature value (such as MFCC) extracted from a single frame.”. Also see para [0034]: “It is also possible to use a vector obtained by combining a plurality of a single-frame feature values. In this case, the feature vector x(t) at time t is expressed by:z(t)=[x i(t), . . . , x N(t)]T (3)x(t)=[z(t−Z)T , . . . , z(t−1)T , z(t)T , z(t+1)T , . . . , z(t+Z)T]T (4)where z(t) is the MFCC at time t; and Z is the number of frames that are used in combining both before and after the frame corresponding to time t.”).
Yamamoto does not exclusively teach the use of series data with likelihood ratios.
Kato, however, analogously teaches series data with likelihood ratios (see col 7 lines 32-38: “A likelihood calculation unit 153 calculates an acoustic likelihood by matching time series data of acoustic feature parameters against a lexical tree stored in a second database 20 and an acoustic model stored in a third database 21 in the self-transition and LR transition to determine an accumulated likelihood by accumulating the acoustic likelihood in a time direction..”)
Before the effective filing date of the claimed invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Yamamoto, Kato, Hunter, and Varin before him or her, to modify the system of claim 8 to include attributes of series data with likelihood ratios in order to operate on ordered data that has regular intervals (see col 7 lines 19-22: “ Acoustic feature parameters are a feature vector obtained by analyzing an input speech at regular intervals (for example, 10 ms; hereinafter, denoted as frames). Therefore, the audio signal is converted into a time series”).
Yamamoto does not explicitly teach wherein the at least one processor is configured to execute the instructions to calculate a new integrated likelihood ratio by using the individual likelihood ratio that is calculated on the basis of the newly obtained element and the integrated likelihood ratio calculated in the past.
Varin, however, analogously teaches to calculate a new integrated likelihood ratio by using the individual likelihood ratio that is calculated on the basis of the newly obtained element and the integrated likelihood ratio calculated in the past (see pg. 2368 section 3: “When we consider an AOP(1) model, in order to compute the first-order pairwise likelihood, we only require the calculation of the following bivariate joint probabilities
PNG
media_image2.png
44
728
media_image2.png
Greyscale
. Whenever , follows a bivariate normal distribution with mean vector
PNG
media_image3.png
46
349
media_image3.png
Greyscale
variances σ^2/(1-ɣ^2)and correlation . Computing the first-order pairwise likelihood requires the approximation (n – 1) of bivariate Gaussian integrals instead of an often prohibitive single n-dimensional Gaussian integral, as in ordinary likelihood inference.”).
Before the effective filing date of the claimed invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Yamamoto, Kato, Hunter, and Varin before him or her, to modify the system of claim 8 to include attributes of wherein the likelihood ratio is an integrated likelihood ratio that is calculated by taking into account a plurality of individual likelihood ratios that are calculated on the basis of two consecutive elements included in the series data in order to not be as prohibitive as ordinary likelihood inferences (see pg. 2368 section 3 : “ Computing the first-order pairwise likelihood requires the approximation of bivariate Gaussian integrals instead of an often prohibitive single n-dimensional Gaussian integral, as in ordinary likelihood inference.”).
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 Andrew A Bracero whose telephone number is (571)270-0592. The examiner can normally be reached Monday - Friday 9:00 a.m. - 5:00 p.m. ET.
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, David Yi can be reached Monday - Friday 9:00 a.m. - 5:00 p.m. ET at (571)
270-7519. 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.
/ANDREW BRACERO/Examiner, Art Unit 2126
/DAVID YI/Supervisory Patent Examiner, Art Unit 2126
Prosecution Insights