检索结果-内蒙古大学图书馆

PSVM: a preference-enhanced SVM model using preference data for classification

science China(Information sciences) 2017年第12期60卷 165-178页

作者： Lerong MA Dandan SONG Lejian LIAO Jingang WANG Engineering Research Center of High Volume Language Information Processing and Cloud Computing Applications School of Computer Science and Technology Beijing Institute of Technology College of Mathematics and Computer Science Yan'an University Search Business Department Alibaba Group

Classification is an essential task in data mining, machine learning and pattern recognition *** classification models focus on distinctive samples from different categories. There are fine-grained differences between data instances within a particular category. These differences form the preference information that is essential for human learning, and, in our view, could also be helpful for classification models. In this paper, we propose a preference-enhanced support vector machine(PSVM), that incorporates preference-pair data as a specific type of supplementary information into SVM. Additionally, we propose a two-layer heuristic sampling method to obtain effective preference-pairs, and an extended sequential minimal optimization(SMO)algorithm to fit PSVM. To evaluate our model, we use the task of knowledge base acceleration-cumulative citation recommendation(KBA-CCR) on the TREC-KBA-2012 dataset and seven other datasets from UCI,Stat Lib and ***. The experimental results show that our proposed PSVM exhibits high performance with official evaluation metrics.

关键词： preference SVM classification sampling sequential minimal optimization(SMO)

来源：评论

学校读者我要写书评

暂无评论

A novel modular wireless sensor networks approach for security applications

引用

International Journal of Security and Networks 2017年第1期12卷 40-50页

作者： Charalampidou, Maria Pavlidis, George Mouroutsos, Spyridon G. Department of Electrical and Computer Engineering Democritus University of Thrace Xanthi67-100 Greece Institute for Language and Speech Processing Athena Research Center Xanthi67-100 Greece

Nowadays surveillance systems are becoming increasingly complex by combining a variety of sensors and systems in order to deliver more accurate decisions. This is due to the fact that the development of a simple and yet efficient algorithm for intruder detection is of a high interest both to academia and the market. This paper presents a novel approach towards the development of a sophisticated decision-making system for accurate intrusion detection. The method is based on three key elements: the assessment of the certainty of the inputs, the quantisation of the inputs using the three-valued logic and the time-based filtering of the sequence of alarms. The algorithm was applied to a wireless sensor network (WSN) that is used for intruder detection of an open area. The overall setup, the theoretical analysis and preliminary evaluation results of the proposed method show that this is an interesting and rather promising approach. Copyright © 2017 Inderscience Enterprises Ltd.

关键词： Wireless sensor networks

来源：评论

学校读者我要写书评

暂无评论

Zipporah: A fast and scalable data cleaning system for noisy web-crawled parallel corpora

Zipporah: A fast and scalable data cleaning system for noisy...

引用

2017 Conference on Empirical Methods in Natural language processing, EMNLP 2017

作者： Xu, Hainan Koehn, Philipp Department of Compute Science Center for Language and Speech Processing Johns Hopkins University 21218 United States

ISBN: (纸本)9781945626838

We introduce Zipporah, a fast and scalable data cleaning system. We propose a novel type of bag-of-words translation feature, and train logistic regression models to classify good data and synthetic noisy data in the proposed feature space. The trained model is used to score parallel sentences in the data pool for selection. As shown in experiments, Zipporah selects a high-quality parallel corpus from a large, mixed quality data pool. In particular, for one noisy dataset, Zipporah achieves a 2.1 BLEU score improvement with using 1/5 of the data over using the entire corpus. © 2017 Association for Computational Linguistics.

关键词： Web crawler

来源：评论

学校读者我要写书评

暂无评论

A potential neurophysiological correlate of electric-acoustic pitch matching in adult cochlear implant users: Pilot data

引用

Cochlear Implants International 2018年第4期19卷 198-209页

作者： Tan, Chin-Tuan Martin, Brett A. Svirsky, Mario A. Department of Electrical and Computer Engineering School of Behavioral and Brain Science (Callier Center for Communication Disorders) University of Texas at Dallas Richardson TX United States Program in Speech-Language-Hearing Sciences and Program in Audiology Graduate Center City University of New York New York NY United States Department of Otolaryngology New York University New York NY United States

The overall goal of this study was to identify an objective physiological correlate of electric-acoustic pitch matching in unilaterally implanted cochlear implant (CI) participants with residual hearing in the non-implanted ear. Electrical and acoustic stimuli were presented in a continuously alternating fashion across ears. The acoustic stimulus and the electrical stimulus were either matched or mismatched in pitch. Auditory evoked potentials were obtained from nine CI users. Results indicated that N1 latency was stimulus-dependent, decreasing when the acoustic frequency of the tone presented to the non-implanted ear was increased. More importantly, there was an additional decrease in N1 latency in the pitch-matched condition. These results indicate the potential utility of N1 latency as an index of pitch matching in CI users. © 2018 Informa UK Limited, trading as Taylor & Francis Group.

关键词： Auditory evoked potential Electric-acoustic pitch matching N1 latency

来源：评论

学校读者我要写书评

暂无评论

Deep Factorization for speech Signal

Deep Factorization for Speech Signal

引用

IEEE International Conference on Acoustics, speech and Signal processing

作者： Lantian Li Dong Wang Yixiang Chen Ying Shi Zhiyuan Tang Thomas Fang Zheng Center for Speech and Language Technologies Research Institute of Information Technology Department of Computer Science and Technology Tsinghua University Beijing 100084 China

ISBN: (纸本)9781538646595

Various informative factors mixed in speech signals, leading to great difficulty when decoding any of the factors. An intuitive idea is to factorize each speech frame into individual informative factors, though it turns out to be highly difficult. Recently, we found that speaker traits, which were assumed to be long-term distributional properties, are actually short-time patterns, and can be learned by a carefully designed deep neural network (DNN). This discovery motivated a cascade deep factorization (CDF) framework that will be presented in this paper. The proposed framework infers speech factors in a sequential way, where factors previously inferred are used as conditional variables when inferring other factors. We will show that this approach can effectively factorize speech signals, and using these factors, the original speech spectrum can be recovered with a high accuracy. This factorization and reconstruction approach provides potential values for many speech processing tasks, e.g., speaker recognition and emotion recognition, as will be demonstrated in the paper.

关键词： speech signal processing speech recognition speaker recognition emotion recognition speaker recognition emotion recognition Voice Signal speech recognition speech factorization signals

来源：评论

学校读者我要写书评

暂无评论

Full-Info Training for Deep Speaker Feature Learning

Full-Info Training for Deep Speaker Feature Learning

引用

IEEE International Conference on Acoustics, speech and Signal processing

作者： Lantian Li Zhiyuan Tang Dong Wang Thomas Fang Zheng Center for Speech and Language Technologies Research Institute of Information Technology Department of Computer Science and Technology Tsinghua University Beijing 100084 China

ISBN: (纸本)9781538646595

In recent studies, it has shown that speaker patterns can be learned from very short speech segments (e.g., 0.3 seconds) by a carefully designed convolutional & time-delay deep neural network (CT-DNN) model. By enforcing the model to discriminate the speakers in the training data, frame-level speaker features can be derived from the last hidden layer. In spite of its good performance, a potential problem of the present model is that it involves a parametric classifier, i.e., the last affine layer, which may consume some discriminative knowledge, thus leading to 'information leak' for the feature learning. This paper presents a full-info training approach that discards the parametric classifier and enforces all the discriminative knowledge learned by the feature net. Our experiments on the Fisher database demonstrate that this new training scheme can produce more coherent features, leading to consistent and notable performance improvement on the speaker verification task.

关键词： speaker recognition deep neural network speaker feature learning speaker recognition Loudspeakers performance boost Training Classifiers Learning

来源：评论

学校读者我要写书评

暂无评论

Deep factorization for speech signal

arXiv

引用

arXiv 2018年

作者： Li, Lantian Wang, Dong Chen, Yixiang Shi, Ying Tang, Zhiyuan Zheng, Thomas Fang Center for Speech and Language Technologies Research Institute of Information Technology Department of Computer Science and Technology Tsinghua University Beijing100084 China

关键词： Factorization

来源：评论

学校读者我要写书评

暂无评论

Automatic speech Recognition for VoIP with Packet Loss Concealment

引用

Procedia computer science 2018年 128卷 72-78页

作者： Adil Bakri Abderrahmane Amrouche Mourad Abbas Lallouani Bouchakour Speech Communication and Signal Processing Laboratory LCPTS Faculty of Electronics and Computer Science USTHB B.P. 32 16111 Bab-Ezzouar Algiers Algeria Scientific and Technical Research Center for the Development of Arabic Language CRSTDLA Algiers Algeria

This paper proposes a packet loss concealment (PLC) technique for increase the robustness of automatic speech recognition (ASR) of speech coded with the G729 codec, on the Voice over Internet Protocol (VoIP). Many of the standard ITU-T CELP based speech coders, such as the G.723.1, G.728, and G.729, model speech reproduction in their decoders. These decoders have enough state information to integrate PLC algorithms directly in the decoder, and are specified as part of their standards in particular by PLC based ITU-T G711 Appendix I. speech is transmitted with source and channel codes optimized, this channel is simulated by two states Markov model to modeled loss packets. The objective of PLC based ITU-T G711 Appendix I is to generate a synthetic speech signal to cover missing data or loss packets in a received bit stream for the ASR application, i.e., to minimize word error rate.

关键词： VoIP RAP OLA PLC G729 ITU-I G711 Appendix I HMM

来源：评论

学校读者我要写书评

暂无评论

On the evaluation of semantic phenomena in neural machine translation using natural language inference

arXiv

引用

arXiv 2018年

作者： Poliak, Adam Belinkov, Yonatan Glass, James van Durme, Benjamin Center for Language and Speech Processing Johns Hopkins University BaltimoreMD21218 United States Computer Science and Artificial Intelligence Laboratory Massachusetts Institute of Technology CambridgeMA02139 United States

We propose a process for investigating the extent to which sentence representations arising from neural machine translation (NMT) systems encode distinct semantic phenomena. We use these representations as features to train a natural language inference (NLI) classifier based on datasets recast from existing semantic annotations. In applying this process to a representative NMT system, we find its encoder appears most suited to supporting inferences at the syntax-semantics interface, as compared to anaphora resolution requiring world-knowledge. We conclude with a discussion on the merits and potential deficiencies of the existing process, and how it may be improved and extended as a broader framework for evaluating semantic coverage.1. Copyright © 2018, The Authors. All rights reserved.

关键词： Neural machine translation

来源：评论

学校读者我要写书评

暂无评论

MCE 2018: The 1st Multi-target speaker detection and identification Challenge Evaluation (MCE) Plan, Dataset and Baseline System

arXiv

引用

arXiv 2018年

作者： Shon, Suwon Dehak, Najim Reynolds, Douglas Glass, James MIT Computer Science and Artificial Intelligence Laboratory CambridgeMA United States Center for Language and Speech Processing Johns Hopkins University Baltimore United States MIT Lincoln Laboratory LexingtonMA United States

The Multitarget Challenge aims to assess how well current speech technology is able to determine whether or not a recorded utterance was spoken by one of a large number of "blacklisted" speakers. It is a form of multi-target speaker detection based on real-world telephone conversations. Data recordings are generated from call center customer-agent conversations. Each conversation is represented by a single i-vector [1]. Given a pool of training and development data from non-Blacklist and Blacklist speakers, the task is to measure how accurately one can detect 1) whether a test recording is spoken by a Blacklist speaker, and 2) which specific Blacklist speaker was talking. Although the primary task will restrict participants to the provided data, participants are allowed to submit secondary systems that use additional data in order to achieve better performance. Copyright © 2018, The Authors. All rights reserved.

关键词： speech recognition

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：