The Experts below are selected from a list of 324 Experts worldwide ranked by ideXlab platform
Zeungnam Bien - One of the best experts on this subject based on the ideXlab platform.
-
IRI - Agglomerative Fuzzy Clustering based on Bayesian Interpretation
2007 IEEE International Conference on Information Reuse and Integration, 2007Co-Authors: Zeungnam BienAbstract:This paper presents iterative Bayesian fuzzy clustering (IBFC), which is based on incorporating integrated adaptive fuzzy clustering (IAFC) with Bayesian decision theory, and finally derives agglomerative IBFC based on its Bayesian Interpretation. IAFC performs a vigilance test so that outliers can be eliminated from learning procedure. However, we have no theoretical background on the rationality of the test. Thus, we claim that the decision and vigilance test of IBFC follow Bayesian minimum risk classification rule within a framework of Bayesian decision theory. Moreover, based on this Interpretation, we propose Agglomerative IBFC capable of clustering data of complex structure. Test on synthetic data shows an outstanding success rate, and test on benchmark data shows that our proposed method performs better than several existing methods.
-
Agglomerative Fuzzy Clustering based on Bayesian Interpretation
2007 IEEE International Conference on Information Reuse and Integration, 2007Co-Authors: Zeungnam BienAbstract:This paper presents iterative Bayesian fuzzy clustering (IBFC), which is based on incorporating integrated adaptive fuzzy clustering (IAFC) with Bayesian decision theory, and finally derives agglomerative IBFC based on its Bayesian Interpretation. IAFC performs a vigilance test so that outliers can be eliminated from learning procedure. However, we have no theoretical background on the rationality of the test. Thus, we claim that the decision and vigilance test of IBFC follow Bayesian minimum risk classification rule within a framework of Bayesian decision theory. Moreover, based on this Interpretation, we propose Agglomerative IBFC capable of clustering data of complex structure. Test on synthetic data shows an outstanding success rate, and test on benchmark data shows that our proposed method performs better than several existing methods.
-
Bayesian Interpretation of Adaptive Fuzzy Neural Network Model
2006 IEEE International Conference on Fuzzy Systems, 2006Co-Authors: Zeungnam BienAbstract:This paper conveys Bayesian Interpretation of improved integrated adaptive fuzzy clustering(IAFC), which is one of the adaptive fuzzy neural network models and suggests upper bound of vigilance parameter, which gives us a guideline to endow IAFC with flexibility within the framework of minimum risk classifier. Besides, we proposed the off-line and on-line learning strategy of IAFC. The proposed techniques are applied to construct facial expression recognition system dealing with neutral, happy, sad, and angry. We empirically show that proposed methods are able to outperform the conventional IAFC.
-
FUZZ-IEEE - Bayesian Interpretation of Adaptive Fuzzy Neural Network Model
2006 IEEE International Conference on Fuzzy Systems, 2006Co-Authors: Sang Wan Lee, Dae-jin Kim, Yong-soo Kim, Zeungnam BienAbstract:This paper conveys Bayesian Interpretation of improved integrated adaptive fuzzy clustering(IAFC), which is one of the adaptive fuzzy neural network models and suggests upper bound of vigilance parameter, which gives us a guideline to endow IAFC with flexibility within the framework of minimum risk classifier. Besides, we proposed the off-line and on-line learning strategy of IAFC. The proposed techniques are applied to construct facial expression recognition system dealing with neutral, happy, sad, and angry. We empirically show that proposed methods are able to outperform the conventional IAFC.
Steve Renals - One of the best experts on this subject based on the ideXlab platform.
-
Hierarchical Bayesian language models for conversational speech recognition
IEEE Transactions on Audio Speech and Language Processing, 2010Co-Authors: Songfang Huang, Steve RenalsAbstract:Traditional n-gram language models are widely used in state-of-the-art large vocabulary speech recognition systems. This simple model suffers from some limitations, such as overfitting of maximum-likelihood estimation and the lack of rich contextual knowledge sources. In this paper, we exploit a hierarchical Bayesian Interpretation for language modeling, based on a nonparametric prior called the Pitman-Yor process. This offers a principled approach to language model smoothing, embedding the power-law distribution for natural language. Experiments on the recognition of conversational speech in multiparty meetings demonstrate that by using hierarchical Bayesian language models, we are able to achieve significant reductions in perplexity and word error rate.
-
Hierarchical Bayesian Language Models for Conversational Speech Recognition
IEEE Transactions on Audio Speech and Language Processing, 2010Co-Authors: Songfang Huang, Steve RenalsAbstract:Traditional n -gram language models are widely used in state-of-the-art large vocabulary speech recognition systems. This simple model suffers from some limitations, such as overfitting of maximum-likelihood estimation and the lack of rich contextual knowledge sources. In this paper, we exploit a hierarchical Bayesian Interpretation for language modeling, based on a nonparametric prior called Pitman-Yor process. This offers a principled approach to language model smoothing, embedding the power-law distribution for natural language. Experiments on the recognition of conversational speech in multiparty meetings demonstrate that by using hierarchical Bayesian language models, we are able to achieve significant reductions in perplexity and word error rate.
Dennis Mcnevin - One of the best experts on this subject based on the ideXlab platform.
-
Response to: Biedermann & Hicks (2019), Commentary on “Dennis McNevin, Bayesian Interpretation of discrete class characteristics, Forensic Science International, 292 (2018) 125–130”
Forensic Science International, 2019Co-Authors: Dennis McnevinAbstract:This letter is a response to the commentary by Biedermann & Hicks (2019) on “Dennis McNevin, Bayesian Interpretation of discrete class characteristics, Forensic Science International, 292 (2018) 125–130”.
-
Bayesian Interpretation of discrete class characteristics
Forensic Science International, 2018Co-Authors: Dennis McnevinAbstract:Abstract Bayesian Interpretation of forensic evidence has become dominated by the likelihood ratio (LR) with a large LR generally considered favourable to the prosecution hypothesis, H P , over the defence hypothesis, H D . However, the LR simply quantifies by how much the prior odds ratio of the probability of H P relative to H D has been improved by the forensic evidence to provide a posterior ratio. Because the prior ratio is mostly neglected, the posterior ratio is largely unknown, regardless of the LR used to improve it. In fact, we show that the posterior ratio will only favour H P when LR is at least as large as the number of things that could possibly be the source of that evidence, all being equally able to contribute. This restriction severely limits the value of evidence to the prosecution when only a single, discrete class characteristic is used to match a subset of these things to the evidence. The limitation can be overcome by examining more than one individual characteristic, as long as they are independent of each other, as they are for the genotypes at multiple loci combined for DNA evidence. We present a criterion for determining how many such characteristics are required. Finally, we conclude that a frequentist Interpretation is inappropriate as a measure of the strength of forensic evidence precisely because it only estimates the denominator of the LR.
Songfang Huang - One of the best experts on this subject based on the ideXlab platform.
-
Hierarchical Bayesian language models for conversational speech recognition
IEEE Transactions on Audio Speech and Language Processing, 2010Co-Authors: Songfang Huang, Steve RenalsAbstract:Traditional n-gram language models are widely used in state-of-the-art large vocabulary speech recognition systems. This simple model suffers from some limitations, such as overfitting of maximum-likelihood estimation and the lack of rich contextual knowledge sources. In this paper, we exploit a hierarchical Bayesian Interpretation for language modeling, based on a nonparametric prior called the Pitman-Yor process. This offers a principled approach to language model smoothing, embedding the power-law distribution for natural language. Experiments on the recognition of conversational speech in multiparty meetings demonstrate that by using hierarchical Bayesian language models, we are able to achieve significant reductions in perplexity and word error rate.
-
Hierarchical Bayesian Language Models for Conversational Speech Recognition
IEEE Transactions on Audio Speech and Language Processing, 2010Co-Authors: Songfang Huang, Steve RenalsAbstract:Traditional n -gram language models are widely used in state-of-the-art large vocabulary speech recognition systems. This simple model suffers from some limitations, such as overfitting of maximum-likelihood estimation and the lack of rich contextual knowledge sources. In this paper, we exploit a hierarchical Bayesian Interpretation for language modeling, based on a nonparametric prior called Pitman-Yor process. This offers a principled approach to language model smoothing, embedding the power-law distribution for natural language. Experiments on the recognition of conversational speech in multiparty meetings demonstrate that by using hierarchical Bayesian language models, we are able to achieve significant reductions in perplexity and word error rate.
Andrzej Drygajlo - One of the best experts on this subject based on the ideXlab platform.
-
Handbook of Biometrics for Forensic Science - Biometric Evidence in Forensic Automatic Speaker Recognition
Handbook of Biometrics for Forensic Science, 2020Co-Authors: Andrzej Drygajlo, Rudolf HaraksimAbstract:The goal of this chapter is to provide a methodology for calculation and Interpretation of biometric evidence in forensic automatic speaker recognition (FASR). It defines processing chains for observed biometric evidence of speech (univariate and multivariate) and for calculating a likelihood ratio as the strength of evidence in the Bayesian Interpretation framework. The calculation of the strength of evidence depends on the speaker models and the similarity scoring used. A processing chain chosen for this purpose is in the close relation with the hypotheses defined in the Bayesian Interpretation framework. Several processing chains are proposed corresponding to the scoring and direct method, which involve univariate and multivariate speech evidence, respectively. This chapter also establishes a methodology to evaluate performance of a chosen FASR method under operating conditions of casework.
-
INTERSPEECH - Statistical methods and Bayesian Interpretation of evidence in forensic automatic speaker recognition.
2020Co-Authors: Andrzej Drygajlo, Didier Meuwly, Anil AlexanderAbstract:The goal of this paper is to establish a robust methodology for forensic automatic speaker recognition (FASR) based on sound statistical and probabilistic methods, and validated using databases recorded in real-life conditions. The Interpretation of recorded speech as evidence in the forensic context presents particular challenges. The means proposed for dealing with them is through Bayesian inference and corpus based methodology. A probabilistic model – the odds form of Bayes’ theorem and likelihood ratio – seems to be an adequate tool for assisting forensic experts in the speaker recognition domain to interpret this evidence. In forensic speaker recognition, statistical modelling techniques are based on the distribution of various features pertaining to the suspect's speech and its comparison to the distribution of the same features in a reference population with respect to the questioned recording. In this paper, the state-of-the-art automatic, text-independent speaker recognition system, using Gaussian mixture model (GMM), is adapted to the Bayesian Interpretation (BI) framework to estimate the within-source variability of the suspected speaker and the between-sources variability, given the questioned recording. This doublestatistical approach (BI-GMM) gives an adequate solution for the Interpretation of the recorded speech as evidence in the judicial process.
-
Wiley Encyclopedia of Forensic Science - Voice: Biometric Analysis and Interpretation of
Wiley Encyclopedia of Forensic Science, 2011Co-Authors: Andrzej DrygajloAbstract:Forensic speaker recognition (FSR) is the process of determining if a specific individual (suspected speaker) could be the source of a questioned voice recording (trace). Results of FSR-based investigations may be of pivotal importance at any stage of an inquiry. The expert's role is to testify to the worth of the voice evidence by using, if possible, a quantitative measure of this worth. This article aims at presenting advances in European research on forensic automatic speaker recognition (FASR), including data-driven tools and related methodology that provide a coherent way of quantifying and presenting recorded voice as biometric evidence, as well as the assessment of its strength in the Bayesian Interpretation framework, compatible with Interpretations in other forensic disciplines. FSR has proved an effective tool, yet there is a constant need for more research because of the difficulties involved in real-life environment (within-speaker variability, between-speaker variability, ambient noise, other sounds, transmission channels, and mismatched recording conditions). Keywords: automatic speaker recognition; voice; biometric evidence; likelihood ratio; strength of biometric evidence; Bayesian Interpretation; questioned recording; speaker features; speaker model; suspected speaker
-
Automatic Speaker Recognition for Forensic Case Assessment and Interpretation
Forensic Speaker Recognition, 2011Co-Authors: Andrzej DrygajloAbstract:Forensic speaker recognition (FSR) is the process of determining if a specific individual (suspected speaker) is the source of a questioned voice recording (trace). The forensic expert’s role is to testify to the worth of the voice evidence by using, if possible, a quantitative measure of this worth. It is up to the judge and/or the jury to use this information as an aid to their deliberations and decision. This chapter aims at presenting research advances in forensic automatic speaker recognition (FASR), including data-driven tools and related methodology, that provide a coherent way of quantifying and presenting recorded voice as biometric evidence , as well as the assessment of its strength (likelihood ratio) in the Bayesian Interpretation framework, compatible with Interpretations in other forensic disciplines. Step-by-step guidelines for the calculation of the biometric evidence and its strength under operating conditions of the casework are provided in this chapter. It also reports on the European Network of Forensic Science Institutes (ENFSI) evaluation campaign through a fake (simulated) case, organized by the Netherlands Forensic Institute (NFI), as an example, where an automatic method using the Gaussian mixture models (GMMs) and the Bayesian Interpretation (BI) framework were implemented for the forensic speaker recognition task.
-
statistical methods and Bayesian Interpretation of evidence in forensic automatic speaker recognition
Conference of the International Speech Communication Association, 2003Co-Authors: Andrzej Drygajlo, Didier Meuwly, Anil AlexanderAbstract:The goal of this paper is to establish a robust methodology for forensic automatic speaker recognition (FASR) based on sound statistical and probabilistic methods, and validated using databases recorded in real-life conditions. The Interpretation of recorded speech as evidence in the forensic context presents particular challenges. The means proposed for dealing with them is through Bayesian inference and corpus based methodology. A probabilistic model – the odds form of Bayes’ theorem and likelihood ratio – seems to be an adequate tool for assisting forensic experts in the speaker recognition domain to interpret this evidence. In forensic speaker recognition, statistical modelling techniques are based on the distribution of various features pertaining to the suspect's speech and its comparison to the distribution of the same features in a reference population with respect to the questioned recording. In this paper, the state-of-the-art automatic, text-independent speaker recognition system, using Gaussian mixture model (GMM), is adapted to the Bayesian Interpretation (BI) framework to estimate the within-source variability of the suspected speaker and the between-sources variability, given the questioned recording. This doublestatistical approach (BI-GMM) gives an adequate solution for the Interpretation of the recorded speech as evidence in the judicial process.