检索结果-南通市图书馆

Voice activity detection based on deep belief networks using likelihood ratio

在线全文

学校读者我要写书评

暂无评论

Journal of Central South University 2016年第1期23卷 145-149页

作者： KIM Sang-Kyun PARK Young-Jin LEE Sangmin Department of Electronic Engineering Inha University Incheon 402-751 Korea Korea Electrotechnology Research Institute (KERI) 111 Hanggaul ro Sangrok Gu An-San shi Kyunggi Do 426-170 Korea Institute for Information and Electronics Research Inha University Incheon 402-751 Korea

A novel technique is proposed to improve the performance of voice activity detection(VAD) by using deep belief networks(DBN) with a likelihood ratio(LR). The likelihood ratio is derived from the speech and noise spectral components that are assumed to follow the Gaussian probability density function(PDF). The proposed algorithm employs DBN learning in order to classify voice activity by using the input signal to calculate the likelihood ratio. Experiments show that the proposed algorithm yields improved results in various noise environments, compared to the conventional VAD algorithms. Furthermore, the DBN based algorithm decreases the detection probability of error with [0.7, 2.6] compared to the support vector machine based algorithm.

关键词： voice activity detection likelihood ratio deep belief networks

BAYESIAN ANALYSIS, likelihood ratio AND INFORMATION

在线全文

学校读者我要写书评

暂无评论

Applied Mathematics(A Journal of Chinese Universities) 1994年第2期9卷 105-108页

作者：张尧庭杜劲松 1. Department of Statistics Wuhan University 430072 Wuhan

We present a new scheme for uncertain reasoning based on Bayes' formula in likelihood ratio form in this paper. When it is used to combine several pieces of uncertan information, the scheme deserves the following meri... 详细信息

关键词： 62A15 68T30 Bayesian analysis likelihood ratio knowledge representation uncertain reasoning

Empirical likelihood ratio-Based Goodness-of-Fit Test for the Laplace Distribution

在线全文

学校读者我要写书评

暂无评论

Communications in Mathematics and Statistics 2016年第4期4卷 459-471页

作者： Hadi Alizadeh Noughabi Department of Statistics University of BirjandBirjandIran

The Laplace distribution can be compared against the normal *** Laplace distribution has an unusual,symmetric shape with a sharp peak and tailsthat are longer than the tails of a normal *** has recently become quitepopular in modeling financial variables(Brownian Laplace motion)like stock returnsbecause of the greater *** Laplace distribution is very extensively reviewed in themonograph(Kotz et *** the laplace distribution and generalizations-a revisit withapplications to communications,economics,engineering,and ***,Boston,2001).In this article,we propose a density-based empirical likelihood ratio(DBELR)goodness-of-fit test statistic for the Laplace *** test statisticis constructed based on the approach proposed by Vexler and Gurevich(Comput StatData Anal 54:531-545,2010).In order to compute the test statistic,parameters of theLaplace distribution are estimated by the maximum likelihood *** valuesand power values of the proposed test are obtained by Monte Carlo ***,power comparisons of the proposed test with some known competing tests are ***,two illustrative examples are presented and analyzed.

关键词： likelihood ratio Laplace distribution Density-based empiricallikelihood ratio Goodness-of-fit test Monte Carlo simulation Power study

Forensic Automatic Speaker Recognition Based on likelihood ratio Using Acoustic-phonetic Features Measured Automatically

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Journal of Forensic Science and Medicine 2015年第2期1卷 119-123页

作者： Huapeng Wang Cuiling Zhang Department of Forensic Science and Technology National Police University of ChinaShenyangLiaoning School of Criminal Investigation Southwest University of Political Science and LawNo.301Baoshe叩AveYubei DistrictChongqingChina

Forensic speaker recognition is experiencing a remarkable paradigm shift in terms of the evaluation framework and presentation of voice *** paper proposes a new method of forensic automatic speaker recognition using the likelihood ratio framework to quantify the strength of voice *** proposed method uses a reference database to calculate the within-and between-speaker *** acoustic-phonetic features are extracted automatically using the software *** effectiveness of the approach was tested using two Mandarin databases:A mobile telephone database and a landline *** experimenfs results indicate that these acoustic-phonetic features do have some discriminating potential and are worth trying in *** automatic acoustic-phonetic features have acceptable discriminative performance and can provide more reliable results in evidence analysis when fused with other kind of voice features.

关键词： Acoustic-phonetic speaker recognition evidence evaluation forensic speaker recognition likelihood ratio

Extracting terms from clinical records of traditional Chinese medicine

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Frontiers of Medicine 2014年第3期8卷 347-351页

作者： Cungen Cao Meng Sun Shi Wang Institute of Computing Technology Chinese Academy of Sciences Beijing 100190 China Software College of Beihang University Beijing100191 China

Health records of traditional Chinese medicine contain valuable clinical inlormation which can be used for improvement of disease treatment and for medical research. In this paper, we present a practical iterative extraction method for extracting terms from the records. The method is based on a set of extraction rules, the Mesh, and the likelihood ratio technique, and achieved a precision rate of 88.18% and a recall rate of 94.21%.

关键词： term extraction rule-based likelihood ratio

Quantifying likelihood in Digital Forensic Investigations

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Journal of Harbin Institute of Technology(New Series) 2014年第6期21卷 1-4页

作者： Richard E Overill Department of Informatics King's College LondonLondon WC2R 2LSUnited Kingdom

While the conventional forensic scientists routinely validate and express the results of their investigations quantitatively using statistical measures from probability theory,digital forensics examiners rarely if ever do *** this paper,we review some of the quantitative tools and techniques which are available for use in digital forensic investigations,including Bayesian networks,complexity theory,information theory and probability theory,and indicate how they may be used to obtain likelihood ratios or odds ratios for the relative plausibility of alternative explanations for the creation of the recovered digital *** potential benefits of such quantitative measures for modern digital forensics are also outlined.

关键词： likelihood ratio odds ratio Bayesian network power law statistics,probability theory complexity theory information theory

Generalized End-to-End Loss for Forensic Speaker Verification

在线全文

学校读者我要写书评

暂无评论

Journal of Systems Science and Information 2023年第2期11卷 264-276页

作者： Huapeng WANG Fangzhou HE Lianquan WU Key Lab of Forensic Science Ministry of JusticeChina(Academy of Forensic Science)Shanghai 200063ChinaSchool of Public Security Information Technology and IntelligenceCriminal Investigation Police University of ChinaShenyang 110854China School of Public Security Information Technology and Intelligence Criminal Investigation Police University of ChinaShenyang 110854China

In recent years,various speech embedding methods based on deep learning have been proposed and have shown better performance in speaker *** new technologies will inevitably promote the development of forensic speaker *** propose a new forensic speaker verification method based on embeddings trained with loss function called generalized end-to-end(GE2E)***,a long short-term memory(LSTM)based deep neural network(DNN)is trained as the embedding extractor,then the cosine similarity scores between embeddings from same speaker comparison pairs and different speaker comparison pairs are trained to represent within-speaker model and between-speaker model respectively,and finally,the cosine similarity scores between the questioned embeddings and enrolled embeddings are evaluated in the above two models to get the likelihood ratio(LR)*** the subset of LibriSpeech,test-other-500,we achieve a new state of the *** all the same speaker comparison pairs and different speaker comparison pairs get correct results and can provide considerable strong evidence strength for courts.

关键词： speaker verification generalized end-to-end loss LSTM likelihood ratio cosine similarity

Objective evaluation of similarity scores derived by Evofinder^(®) system for marks on bullets fired from Chinese Norinco QSZ-92 pistols

在线全文

学校读者我要写书评

暂无评论

Forensic Sciences Research 2022年第1期7卷 40-46页

作者： Feng Dong Yabin Zhao Yaping Luo Weifang Zhang Yuesong Li School of Forensic Science People’s Public Security University of ChinaBeijingChina Shanghai Key Laboratory of Crime Scene Evidence Shanghai Research Institute of Criminal Science and TechnologyShanghaiChina Graduate School People’s Public Security University of ChinaBeijingChina

In recent years,many studies have been conducted in the field of firearm identification with the objective of providing an objective method of evaluating the comparison of cartridge ***,less attention has been paid to the objective evaluation of bullet *** this study,1000 registered Chinese Norinco QSZ-92 pistols were used,and a database of 2996 bullets was *** the receiver operating characteristic(ROC)curve and the score-based likelihood ratio method were used to objectively evaluate the similarity scores derived by the EvofinderVR *** results indicate that this system has excellent ability to discriminate between the selected *** paper proposes an objective evaluation method,which serves as a response to the ongoing debates regarding the foundation of the discipline.

关键词： Forensic sciences firearm identification EvofinderVR likelihood ratio automatic correlation

Specificity of Characteristic Marks on Cartridge Cases from 3070 Consecutive Firings of a Chinese Norinco QSZ‑929 mm Pistol

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Journal of Forensic Science and Medicine 2019年第2期5卷 87-94页

作者： Feng Dong Yabin Zhao Yaping Luo Weifang Zhang Kaifeng Zhang School of Forensic Science People’s Public Security University of ChinaBeijingChina Shanghai Key Laboratory of Crime Scene Evidence Shanghai Research Institute of Criminal Science and TechnologyShanghaiChina Graduate School People’s Public Security University of ChinaBeijingChina

The specificity of marks was evaluated by differences between marks left by the same firearm(within variability)and differences between marks left by different firearms(between variability).The within variability and between variability of firing pin impressions and breech face marks on cartridge cases from over 3070 consecutive firings of a Chinese Norinco QSZ‑929 mm pistol were investigated based on the automated ballistic identification system Evofinder®.The first 20 cartridge cases and one random specimen out of every set of 10 consecutive cartridge cases of the remaining 3050 cases were entered into the Evofinder®***,325 cartridges fired from the same pistol were used to estimate the within ***,88 cartridges fired from 88 different pistols of the same model(one cartridge each)were used to estimate the between *** study established a database containing 413 *** used the 325 cartridges fired from the same pistol as a correlation baseline in the *** firing pin and breech scores and gaps between these scores were used to analyze the *** the score evaluation process,a likelihood ratio system was used to evaluate the likelihood ratio *** Evofinder®system correctly discriminated the matches,and the likelihood ratios provided strong support for the true state of the *** addition,we found that the score gap was usually greater than the other score differences by more than 1000%in firing pin list and 100%in the breech *** analysis of the score and gap can help examiners use the Evofinder®system more efficiently.

关键词： Automatic correlation Evofinder® firearm identification likelihood ratio