The Experts below are selected from a list of 300 Experts worldwide ranked by ideXlab platform

Noboru Harada - One of the best experts on this subject based on the ideXlab platform.

  • Unsupervised Detection of Anomalous Sound Based on Deep Learning and the Neyman–Pearson Lemma
    IEEE ACM Transactions on Audio Speech and Language Processing, 2019
    Co-Authors: Yuma Koizumi, Shoichiro Saito, Hisashi Uematsu, Yuta Kawachi, Noboru Harada
    Abstract:

    This paper proposes a novel optimization principle and its implementation for unsupervised anomaly detection in sound ADS using an autoencoder AE. The goal of the unsupervised-ADS is to detect unknown anomalous sounds without training data of anomalous sounds. The use of an AE as a normal model is a state-of-the-art technique for the unsupervised-ADS. To decrease the false positive rate FPR, the AE is trained to minimize the reconstruction error of normal sounds, and the anomaly score is calculated as the reconstruction error of the observed sound. Unfortunately, since this training procedure does not take into account the anomaly score for anomalous sounds, the true positive rate TPR does not necessarily increase. In this study, we define an objective function based on the Neyman–Pearson Lemma by considering the ADS as a statistical hypothesis test. The proposed objective function trains the AE to maximize the TPR under an arbitrary low FPR condition. To calculate the TPR in the objective function, we consider that the set of anomalous sounds is the complementary set of normal sounds and simulate anomalous sounds by using a rejection sampling algorithm. Through experiments using synthetic data, we found that the proposed method improved the performance measures of the ADS under low FPR conditions. In addition, we confirmed that the proposed method could detect anomalous sounds in real environments.

  • unsupervised detection of anomalous sound based on deep learning and the neyman Pearson Lemma
    IEEE Transactions on Audio Speech and Language Processing, 2019
    Co-Authors: Yuma Koizumi, Shoichiro Saito, Hisashi Uematsu, Yuta Kawachi, Noboru Harada
    Abstract:

    This paper proposes a novel optimization principle and its implementation for unsupervised anomaly detection in sound ADS using an autoencoder AE. The goal of the unsupervised-ADS is to detect unknown anomalous sounds without training data of anomalous sounds. The use of an AE as a normal model is a state-of-the-art technique for the unsupervised-ADS. To decrease the false positive rate FPR, the AE is trained to minimize the reconstruction error of normal sounds, and the anomaly score is calculated as the reconstruction error of the observed sound. Unfortunately, since this training procedure does not take into account the anomaly score for anomalous sounds, the true positive rate TPR does not necessarily increase. In this study, we define an objective function based on the Neyman–Pearson Lemma by considering the ADS as a statistical hypothesis test. The proposed objective function trains the AE to maximize the TPR under an arbitrary low FPR condition. To calculate the TPR in the objective function, we consider that the set of anomalous sounds is the complementary set of normal sounds and simulate anomalous sounds by using a rejection sampling algorithm. Through experiments using synthetic data, we found that the proposed method improved the performance measures of the ADS under low FPR conditions. In addition, we confirmed that the proposed method could detect anomalous sounds in real environments.

  • unsupervised detection of anomalous sound based on deep learning and the neyman Pearson Lemma
    arXiv: Machine Learning, 2018
    Co-Authors: Yuma Koizumi, Shoichiro Saito, Hisashi Uematsum Yuta Kawachi, Noboru Harada
    Abstract:

    This paper proposes a novel optimization principle and its implementation for unsupervised anomaly detection in sound (ADS) using an autoencoder (AE). The goal of unsupervised-ADS is to detect unknown anomalous sound without training data of anomalous sound. Use of an AE as a normal model is a state-of-the-art technique for unsupervised-ADS. To decrease the false positive rate (FPR), the AE is trained to minimize the reconstruction error of normal sounds and the anomaly score is calculated as the reconstruction error of the observed sound. Unfortunately, since this training procedure does not take into account the anomaly score for anomalous sounds, the true positive rate (TPR) does not necessarily increase. In this study, we define an objective function based on the Neyman-Pearson Lemma by considering ADS as a statistical hypothesis test. The proposed objective function trains the AE to maximize the TPR under an arbitrary low FPR condition. To calculate the TPR in the objective function, we consider that the set of anomalous sounds is the complementary set of normal sounds and simulate anomalous sounds by using a rejection sampling algorithm. Through experiments using synthetic data, we found that the proposed method improved the performance measures of ADS under low FPR conditions. In addition, we confirmed that the proposed method could detect anomalous sounds in real environments.

  • optimizing acoustic feature extractor for anomalous sound detection based on neyman Pearson Lemma
    European Signal Processing Conference, 2017
    Co-Authors: Yuma Koizumi, Shoichiro Saito, Hisashi Uematsu, Noboru Harada
    Abstract:

    We propose a method for optimizing an acoustic feature extractor for anomalous sound detection (ASD). Most ASD systems adopt outlier-detection techniques because it is difficult to collect a massive amount of anomalous sound data. To improve the performance of such outlier-detection-based ASD, it is essential to extract a set of efficient acoustic features that is suitable for identifying anomalous sounds. However, the ideal property of a set of acoustic features that maximizes ASD performance has not been clarified. By considering outlier-detection-based ASD as a statistical hypothesis test, we defined optimality as an objective function that adopts Neyman-Pearson Lemma; the acoustic feature extractor is optimized to extract a set of acoustic features which maximize the true positive rate under an arbitrary false positive rate. The variational auto-encoder is applied as an acoustic feature extractor and optimized to maximize the objective function. We confirmed that the proposed method improved the F-measure score from 0.02 to 0.06 points compared to those of conventional methods, and ASD results of a stereolithography 3D-printer in a real-environment show that the proposed method is effective in identifying anomalous sounds.

  • EUSIPCO - Optimizing acoustic feature extractor for anomalous sound detection based on Neyman-Pearson Lemma
    2017 25th European Signal Processing Conference (EUSIPCO), 2017
    Co-Authors: Yuma Koizumi, Shoichiro Saito, Hisashi Uematsu, Noboru Harada
    Abstract:

    We propose a method for optimizing an acoustic feature extractor for anomalous sound detection (ASD). Most ASD systems adopt outlier-detection techniques because it is difficult to collect a massive amount of anomalous sound data. To improve the performance of such outlier-detection-based ASD, it is essential to extract a set of efficient acoustic features that is suitable for identifying anomalous sounds. However, the ideal property of a set of acoustic features that maximizes ASD performance has not been clarified. By considering outlier-detection-based ASD as a statistical hypothesis test, we defined optimality as an objective function that adopts Neyman-Pearson Lemma; the acoustic feature extractor is optimized to extract a set of acoustic features which maximize the true positive rate under an arbitrary false positive rate. The variational auto-encoder is applied as an acoustic feature extractor and optimized to maximize the objective function. We confirmed that the proposed method improved the F-measure score from 0.02 to 0.06 points compared to those of conventional methods, and ASD results of a stereolithography 3D-printer in a real-environment show that the proposed method is effective in identifying anomalous sounds.

Xun Yu Zhou - One of the best experts on this subject based on the ideXlab platform.

  • a generalized neyman Pearson Lemma for g probabilities
    Probability Theory and Related Fields, 2010
    Co-Authors: Xun Yu Zhou
    Abstract:

    This paper is concerned with hypothesis tests for g-probabilities, a class of nonlinear probability measures. The problem is shown to be a special case of a general stochastic optimization problem where the objective is to choose the terminal state of certain backward stochastic differential equations so as to minimize a g-expectation. The latter is solved with a stochastic maximum principle approach. Neyman–Pearson type results are thereby derived for the original problem with both simple and randomized tests. It turns out that the likelihood ratio in the optimal tests is nothing else than the ratio of the adjoint processes associated with the maximum principle. Concrete examples, ranging from the classical simple tests, financial market modelling with ambiguity, to super- and sub-pricing of contingent claims and to risk measures, are presented to illustrate the applications of the results obtained.

  • A generalized Neyman–Pearson Lemma for g -probabilities
    Probability Theory and Related Fields, 2009
    Co-Authors: Xun Yu Zhou
    Abstract:

    This paper is concerned with hypothesis tests for g-probabilities, a class of nonlinear probability measures. The problem is shown to be a special case of a general stochastic optimization problem where the objective is to choose the terminal state of certain backward stochastic differential equations so as to minimize a g-expectation. The latter is solved with a stochastic maximum principle approach. Neyman–Pearson type results are thereby derived for the original problem with both simple and randomized tests. It turns out that the likelihood ratio in the optimal tests is nothing else than the ratio of the adjoint processes associated with the maximum principle. Concrete examples, ranging from the classical simple tests, financial market modelling with ambiguity, to super- and sub-pricing of contingent claims and to risk measures, are presented to illustrate the applications of the results obtained.

  • A generalized Neyman–Pearson Lemma for g-probabilities
    Probability Theory and Related Fields, 2009
    Co-Authors: Xun Yu Zhou
    Abstract:

    This paper is concerned with hypothesis tests for g-probabilities, a class of nonlinear probability measures. The problem is shown to be a special case of a general stochastic optimization problem where the objective is to choose the terminal state of certain backward stochastic differential equations so as to minimize a g-expectation. The latter is solved with a stochastic maximum principle approach. Neyman-Pearson type results are thereby derived for the original problem with both simple and randomized tests. It turns out that the likelihood ratio in the optimal tests is nothing else than the ratio of the adjoint processes associated with the maximum principle. Concrete examples, ranging from the classical simple tests, financial market modelling with ambiguity, to super- and sub-pricing of contingent claims and to risk measures, are presented to illustrate the applications of the results obtained. © 2009 Springer-Verlag

  • the neyman Pearson Lemma under g probability
    Comptes Rendus Mathematique, 2008
    Co-Authors: Xun Yu Zhou
    Abstract:

    Abstract The Neyman–Pearson fundamental Lemma is generalized under g-probability. With convexity assumptions, a sufficient and necessary condition which characterizes the optimal randomized tests is obtained via a maximum principle for stochastic control. To cite this article: S. Ji, X.Y. Zhou, C. R. Acad. Sci. Paris, Ser. I 346 (2008).

  • The Neyman–Pearson Lemma under g-probability⁎⁎⁎
    Comptes Rendus Mathematique, 2008
    Co-Authors: Xun Yu Zhou
    Abstract:

    Abstract The Neyman–Pearson fundamental Lemma is generalized under g-probability. With convexity assumptions, a sufficient and necessary condition which characterizes the optimal randomized tests is obtained via a maximum principle for stochastic control. To cite this article: S. Ji, X.Y. Zhou, C. R. Acad. Sci. Paris, Ser. I 346 (2008).

Hamzeh Torabi - One of the best experts on this subject based on the ideXlab platform.

  • Cooperative spectrum sensing against noise uncertainty using Neyman-Pearson Lemma on fuzzy hypothesis test
    Applied Soft Computing, 2013
    Co-Authors: Abdolreza Mohammadi, Mohammad Reza Taban, Jamshid Abouei, Hamzeh Torabi
    Abstract:

    In this paper, we consider the problem of cooperative spectrum sensing in the presence of the noise power uncertainty. We propose a new spectrum sensing method based on the fuzzy hypothesis test (FHT) that utilizes membership functions as hypotheses for the modeling and analyzing such uncertainty. In particular, we apply the Neyman-Pearson Lemma on the FHT and propose a threshold-based local detector at each secondary user (SU) in which the threshold depends on the noise power uncertainty. In the proposed scheme, a centralized manner in the cooperative spectrum sensing is deployed in which each SU sends its one bit decision to a fusion center. The fusion center makes a final decision about the absence/presence of a primary user (PU). The performance of the PU's signal detection is evaluated by the probability of signal detection for a specific signal to noise ratio when the probability of false alarm is set to a fixed value. The performance of the proposed algorithm is compared numerically with two classical threshold-based energy detectors. Simulation results show that the proposed algorithm considerably outperforms the methods with a bi-thresholds energy detector and a simple energy detector in the presence of the noise power uncertainty.

  • A Generalized Version of Neyman-Pearson Lemma for Testing Fuzzy Hypotheses Based on r-Level Sets
    Communications in Statistics - Theory and Methods, 2012
    Co-Authors: Hamzeh Torabi
    Abstract:

    In testing statistical hypotheses, as in other statistical problems, we may be confronted with fuzzy concepts. In this article, we first redefine some concepts in testing of fuzzy hypotheses and then introduce a generalized version of Neyman-Pearson Lemma for testing fuzzy hypotheses using r-levels. Finally, two numerical examples are presented to demonstrate the proposed approach.

  • The Most Powerful Tests for Fuzzy Hypotheses Testing with Vague Data
    2009
    Co-Authors: Hamzeh Torabi, S. M. Mirhosseini
    Abstract:

    In hypotheses testing, such as other statistical problems, we may confront imprecise concepts. One case is a situation in which both hypotheses and observations are imprecise. In this paper, we redefine some concepts about fuzzy hypotheses testing, and then we give the Neyman-Pearson Lemma for fuzzy hypotheses testing with fuzzy observations. Finally, we give some applied examples.

  • Neyman–Pearson Lemma for Fuzzy Hypotheses Testing with Vague Data
    Metrika, 2006
    Co-Authors: Hamzeh Torabi, Javad Behboodian, S. Mahmoud Taheri
    Abstract:

    In hypotheses testing, such as other statistical problems, we may confront imprecise concepts. One case is a situation in which both hypotheses and observations are imprecise. This paper tries to develop a new approach for testing fuzzy hypothesis when the available data are fuzzy, too. First, some definitions are provided, such as: fuzzy sample space, fuzzy-valued random sample, and fuzzy-valued random variable. Then, the problem of fuzzy hypothesis testing with vague data is formulated. Finally, we state and prove a generalized Neyman–Pearson Lemma for such problem. The proposed approach is illustrated by some numerical examples.

  • neyman Pearson Lemma for fuzzy hypotheses testing with vague data
    Metrika, 2006
    Co-Authors: Hamzeh Torabi, Javad Behboodian, Mahmoud S Taheri
    Abstract:

    In hypotheses testing, such as other statistical problems, we may confront imprecise concepts. One case is a situation in which both hypotheses and observations are imprecise. This paper tries to develop a new approach for testing fuzzy hypothesis when the available data are fuzzy, too. First, some definitions are provided, such as: fuzzy sample space, fuzzy-valued random sample, and fuzzy-valued random variable. Then, the problem of fuzzy hypothesis testing with vague data is formulated. Finally, we state and prove a generalized Neyman–Pearson Lemma for such problem. The proposed approach is illustrated by some numerical examples.

Yuma Koizumi - One of the best experts on this subject based on the ideXlab platform.

  • Unsupervised Detection of Anomalous Sound Based on Deep Learning and the Neyman–Pearson Lemma
    IEEE ACM Transactions on Audio Speech and Language Processing, 2019
    Co-Authors: Yuma Koizumi, Shoichiro Saito, Hisashi Uematsu, Yuta Kawachi, Noboru Harada
    Abstract:

    This paper proposes a novel optimization principle and its implementation for unsupervised anomaly detection in sound ADS using an autoencoder AE. The goal of the unsupervised-ADS is to detect unknown anomalous sounds without training data of anomalous sounds. The use of an AE as a normal model is a state-of-the-art technique for the unsupervised-ADS. To decrease the false positive rate FPR, the AE is trained to minimize the reconstruction error of normal sounds, and the anomaly score is calculated as the reconstruction error of the observed sound. Unfortunately, since this training procedure does not take into account the anomaly score for anomalous sounds, the true positive rate TPR does not necessarily increase. In this study, we define an objective function based on the Neyman–Pearson Lemma by considering the ADS as a statistical hypothesis test. The proposed objective function trains the AE to maximize the TPR under an arbitrary low FPR condition. To calculate the TPR in the objective function, we consider that the set of anomalous sounds is the complementary set of normal sounds and simulate anomalous sounds by using a rejection sampling algorithm. Through experiments using synthetic data, we found that the proposed method improved the performance measures of the ADS under low FPR conditions. In addition, we confirmed that the proposed method could detect anomalous sounds in real environments.

  • unsupervised detection of anomalous sound based on deep learning and the neyman Pearson Lemma
    IEEE Transactions on Audio Speech and Language Processing, 2019
    Co-Authors: Yuma Koizumi, Shoichiro Saito, Hisashi Uematsu, Yuta Kawachi, Noboru Harada
    Abstract:

    This paper proposes a novel optimization principle and its implementation for unsupervised anomaly detection in sound ADS using an autoencoder AE. The goal of the unsupervised-ADS is to detect unknown anomalous sounds without training data of anomalous sounds. The use of an AE as a normal model is a state-of-the-art technique for the unsupervised-ADS. To decrease the false positive rate FPR, the AE is trained to minimize the reconstruction error of normal sounds, and the anomaly score is calculated as the reconstruction error of the observed sound. Unfortunately, since this training procedure does not take into account the anomaly score for anomalous sounds, the true positive rate TPR does not necessarily increase. In this study, we define an objective function based on the Neyman–Pearson Lemma by considering the ADS as a statistical hypothesis test. The proposed objective function trains the AE to maximize the TPR under an arbitrary low FPR condition. To calculate the TPR in the objective function, we consider that the set of anomalous sounds is the complementary set of normal sounds and simulate anomalous sounds by using a rejection sampling algorithm. Through experiments using synthetic data, we found that the proposed method improved the performance measures of the ADS under low FPR conditions. In addition, we confirmed that the proposed method could detect anomalous sounds in real environments.

  • unsupervised detection of anomalous sound based on deep learning and the neyman Pearson Lemma
    arXiv: Machine Learning, 2018
    Co-Authors: Yuma Koizumi, Shoichiro Saito, Hisashi Uematsum Yuta Kawachi, Noboru Harada
    Abstract:

    This paper proposes a novel optimization principle and its implementation for unsupervised anomaly detection in sound (ADS) using an autoencoder (AE). The goal of unsupervised-ADS is to detect unknown anomalous sound without training data of anomalous sound. Use of an AE as a normal model is a state-of-the-art technique for unsupervised-ADS. To decrease the false positive rate (FPR), the AE is trained to minimize the reconstruction error of normal sounds and the anomaly score is calculated as the reconstruction error of the observed sound. Unfortunately, since this training procedure does not take into account the anomaly score for anomalous sounds, the true positive rate (TPR) does not necessarily increase. In this study, we define an objective function based on the Neyman-Pearson Lemma by considering ADS as a statistical hypothesis test. The proposed objective function trains the AE to maximize the TPR under an arbitrary low FPR condition. To calculate the TPR in the objective function, we consider that the set of anomalous sounds is the complementary set of normal sounds and simulate anomalous sounds by using a rejection sampling algorithm. Through experiments using synthetic data, we found that the proposed method improved the performance measures of ADS under low FPR conditions. In addition, we confirmed that the proposed method could detect anomalous sounds in real environments.

  • optimizing acoustic feature extractor for anomalous sound detection based on neyman Pearson Lemma
    European Signal Processing Conference, 2017
    Co-Authors: Yuma Koizumi, Shoichiro Saito, Hisashi Uematsu, Noboru Harada
    Abstract:

    We propose a method for optimizing an acoustic feature extractor for anomalous sound detection (ASD). Most ASD systems adopt outlier-detection techniques because it is difficult to collect a massive amount of anomalous sound data. To improve the performance of such outlier-detection-based ASD, it is essential to extract a set of efficient acoustic features that is suitable for identifying anomalous sounds. However, the ideal property of a set of acoustic features that maximizes ASD performance has not been clarified. By considering outlier-detection-based ASD as a statistical hypothesis test, we defined optimality as an objective function that adopts Neyman-Pearson Lemma; the acoustic feature extractor is optimized to extract a set of acoustic features which maximize the true positive rate under an arbitrary false positive rate. The variational auto-encoder is applied as an acoustic feature extractor and optimized to maximize the objective function. We confirmed that the proposed method improved the F-measure score from 0.02 to 0.06 points compared to those of conventional methods, and ASD results of a stereolithography 3D-printer in a real-environment show that the proposed method is effective in identifying anomalous sounds.

  • EUSIPCO - Optimizing acoustic feature extractor for anomalous sound detection based on Neyman-Pearson Lemma
    2017 25th European Signal Processing Conference (EUSIPCO), 2017
    Co-Authors: Yuma Koizumi, Shoichiro Saito, Hisashi Uematsu, Noboru Harada
    Abstract:

    We propose a method for optimizing an acoustic feature extractor for anomalous sound detection (ASD). Most ASD systems adopt outlier-detection techniques because it is difficult to collect a massive amount of anomalous sound data. To improve the performance of such outlier-detection-based ASD, it is essential to extract a set of efficient acoustic features that is suitable for identifying anomalous sounds. However, the ideal property of a set of acoustic features that maximizes ASD performance has not been clarified. By considering outlier-detection-based ASD as a statistical hypothesis test, we defined optimality as an objective function that adopts Neyman-Pearson Lemma; the acoustic feature extractor is optimized to extract a set of acoustic features which maximize the true positive rate under an arbitrary false positive rate. The variational auto-encoder is applied as an acoustic feature extractor and optimized to maximize the objective function. We confirmed that the proposed method improved the F-measure score from 0.02 to 0.06 points compared to those of conventional methods, and ASD results of a stereolithography 3D-printer in a real-environment show that the proposed method is effective in identifying anomalous sounds.

Alex Acero - One of the best experts on this subject based on the ideXlab platform.

  • A GENERATIVE-DISCRIMINATIVE FRAMEWORK USING ENSEMBLE METHODS FOR TEXT-DEPENDENT SPEAKER VERIFICATION
    2016
    Co-Authors: Amarnag Subramanyal, Zhengyou Zhang, Arun C. Surendran, Patrick Nguyen, Alex Acero
    Abstract:

    Speaker Verification can be treated as a statistical hypothesis testing problem. The most commonly used approach is the likelihood ratio test (LRT), which can be shown to be optimal using the Neymann-Pearson Lemma. However, in most practical situations the Neymann-Pearson Lemma does not apply. In this paper, we present a more ro-bust approach that makes use of a hybrid generative-discriminative framework for text-dependent speaker verification. Our algorithm makes use of a generative models to learn the characteristics of a speaker and then discriminative models to discriminate between a speaker and an impostor. One of the advantages of the proposed al-gorithm is that it does not require us to retrain the generative model. The proposed model, on an average, yields 36.41 % relative improve-ment in EER over a LRT

  • ICASSP (4) - A Generative-Discriminative Framework using Ensemble Methods for Text-Dependent Speaker Verification
    2007 IEEE International Conference on Acoustics Speech and Signal Processing - ICASSP '07, 2007
    Co-Authors: Amarnag Subramanya, Zhengyou Zhang, Arun C. Surendran, Patrick Nguyen, Mukund Narasimhan, Alex Acero
    Abstract:

    Speaker verification can be treated as a statistical hypothesis testing problem. The most commonly used approach is the likelihood ratio test (LRT), which can be shown to be optimal using the Neymann-Pearson Lemma. However, in most practical situations the Neymann-Pearson Lemma does not apply. In this paper, we present a more robust approach that makes use of a hybrid generative-discriminative framework for text-dependent speaker verification. Our algorithm makes use of a generative models to learn the characteristics of a speaker and then discriminative models to discriminate between a speaker and an impostor. One of the advantages of the proposed algorithm is that it does not require us to retrain the generative model. The proposed model, on an average, yields 36.41% relative improvement in EER over a LRT.