检索结果-内蒙古大学图书馆

IEEE Workshop on Automatic speech Recognition and Understanding

作者： Vijayaditya Peddinti Guoguo Chen Vimal Manohar Tom Ko Daniel Povey Sanjeev Khudanpur Center for language and speech processing The Johns Hopkins University Baltimore MD USA Huawei Noah's Ark Research Lab Hong Kong China Human Language Technology Center of Excellence The Johns Hopkins University Baltimore MD USA

Multi-style training, using data which emulates a variety of possible test scenarios, is a popular approach towards robust acoustic modeling. However acoustic models capable of exploiting large amounts of training data in a comparatively short amount of training time are essential. In this paper we tackle the problem of reverberant speech recognition using 5500 hours of simulated reverberant data. We use time-delay neural network (TDNN) architecture, which is capable of tackling long-term interactions between speech and corrupting sources in reverberant environments. By sub-sampling the outputs at TDNN layers across time steps, training time is substantially reduced. Combining this with distributed-optimization we show that the TDNN can be trained in 3 days using up to 32 GPUs. Further, iVectors are used as an input to the neural network to perform instantaneous speaker and environment adaptation. Finally, recurrent neural network language models are applied to the lattices to further improve the performance. Our system is shown to provide state-of-the-art results in the IARPA ASpIRE challenge, with 26.5% WER on the dev Jest set.

关键词： speech Context Training data Neural networks Training Acoustics Databases

来源：评论

学校读者我要写书评

暂无评论

Efficient elicitation of annotations for human evaluation of machine translation 9

Efficient elicitation of annotations for human evaluation of...

引用

9th Workshop on Statistical Machine Translation, WMT 2014 at the 52nd Conference of the Associationfor Computational Linguistics, ACL 2014

作者： Sakaguchi, Keisuke Post, Matt Van Durme, Benjamin Center for Language and Speech Processing United States Human Language Technology Center of Excellence Johns Hopkins University BaltimoreMD United States

ISBN: (纸本)9781941643174

A main output of the annual Workshop on Statistical Machine Translation (WMT) is a ranking of the systems that participated in its shared translation tasks, produced by aggregating pairwise sentencelevel comparisons collected from human judges. Over the past few years, there have been a number of tweaks to the aggregation formula in attempts to address issues arising from the inherent ambiguity and subjectivity of the task, as well as weaknesses in the proposed models and the manner of model selection. We continue this line of work by adapting the TrueSkillTM algorithm an online approach for modeling the relative skills of players in ongoing competitions, such as Microsoft s Xbox Live to the human evaluation of machine translation output. Our experimental results show that TrueSkill outperforms other recently proposed models on accuracy, and also can significantly reduce the number of pairwise annotations that need to be collected by sampling non-uniformly from the space of system competitions. © Proceedings of the Annual Meeting of the Association for Computational Linguistics 2014.

关键词： Machine translation

来源：评论

学校读者我要写书评

暂无评论

Is the Stanford Dependency Representation Semantic? 2

Is the Stanford Dependency Representation Semantic?

引用

2nd Workshop on EVENTS: Definition, Detection, Coreference, and Representation, EVENTS 2014 at the 52nd Annual Meeting of the Association for Computational Linguistics, ACL 2014

作者： Rudinger, Rachel Van Durme, Benjamin Center for Language and Speech Processing Johns Hopkins University United States Human Language Technology Center of Excellence Johns Hopkins University United States

ISBN: (纸本)9781941643143

The Stanford Dependencies are a deep syntactic representation that are widely used for semantic tasks, like Recognizing Textual Entailment. But do they capture all of the semantic information a meaning representation ought to convey? This paper explores this question by investigating the feasibility of mapping Stanford dependency parses to Hobbsian Logical Form, a practical, event-theoretic semantic representation, using only a set of deterministic rules. Although we find that such a mapping is possible in a large number of cases, we also find cases for which such a mapping seems to require information beyond what the Stanford Dependencies encode. These cases shed light on the kinds of semantic information that are and are not present in the Stanford Dependencies. © 2014 Association for Computational Linguistics

关键词： Mapping

来源：评论

学校读者我要写书评

暂无评论

Augmenting FrameNet Via PPDB 2

Augmenting FrameNet Via PPDB

引用

2nd Workshop on EVENTS: Definition, Detection, Coreference, and Representation, EVENTS 2014 at the 52nd Annual Meeting of the Association for Computational Linguistics, ACL 2014

作者： Rastogi, Pushpendre Van Durme, Benjamin Center for Language and Speech Processing Johns Hopkins University United States Human Language Technology Center of Excellence Johns Hopkins University United States

FrameNet is a lexico-semantic dataset that embodies the theory of frame semantics. Like other semantic databases, FrameNet is incomplete. We augment it via the paraphrase database, PPDB, and gain a threefold increase ...

ISBN: (纸本)9781941643143

关键词：

来源：评论

学校读者我要写书评

暂无评论

Low-resource open vocabulary keyword search using point process models 15

Low-resource open vocabulary keyword search using point proc...

引用

15th Annual Conference of the International speech Communication Association: Celebrating the Diversity of Spoken languages, INTERspeech 2014

作者： Liu, Chunxi Jansen, Aren Chen, Guoguo Kintzley, Keith Trmal, Jan Khudanpur, Sanjeev Center for Language and Speech Processing Department of Electrical and Computer Engineering United States Human Language Technology Center of Excellence Johns Hopkins University BaltimoreMD United States

The point process model (PPM) for keyword search is a whole-word parametric modeling framework based on the timing of phonetic events rather than the evolution of frame-level phonetic likelihoods. Recent progress in PPM training and decoding algorithms has yielded state-of-the-art phonetic search performance in high-resource settings, both in terms of accuracy and computational efficiency. In this paper, we consider PPM application to low-resource settings where the amount of transcribed speech is severely limited and the pronunciation dictionary is incomplete. By using (i) state-of-the-art deep neural network acoustic models to generate phonetic events and (ii) grapheme-to-phoneme conversion to generate pronunciations for out-of-vocabulary (OOV) keywords, we find the PPM system reaches state-of-the-art OOV search performance at a small computational cost. Moreover, due to their complementary methodologies, combining PPM outputs with the LVCSR baseline produces average relative ATWV improvements of 7% and 50% for in-vocabulary and OOV keywords, respectively (16% overall). Copyright © 2014 ISCA.

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Combination of FST and CN search in Spoken Term Detection 15

Combination of FST and CN search in Spoken Term Detection

引用

15th Annual Conference of the International speech Communication Association: Celebrating the Diversity of Spoken languages, INTERspeech 2014

作者： Chiu, Justin Wang, Yun Trmal, Jan Povey, Daniel Chen, Guoguo Rudnicky, Alexander Language Technologies Institute Carnegie Mellon University PittsburghPA United States Center for Language and Speech Processing and Human Language Technology Center of Excellence Johns Hopkins University BaltimoreMD United States

Spoken Term Detection (STD) focuses on finding instances of a particular spoken word or phrase in an audio corpus. Most STD systems have a two-step pipeline, ASR followed by search. Two approaches to search are common, Confusion Network (CN) based search and Finite State Transducer (FST) based search. In this paper, we examine combination of these two different search approaches, using the same ASR output. We find that the CN search performs better on shorter queries, and FST search performs better on longer queries. By combining the different search results from the same ASR decoding, we achieve better performance compared to either search approach on its own. We also find that this improvement is additive to the usual combination of decoder results using different modeling techniques. Copyright © 2014 ISCA.

关键词： Decoding

来源：评论

学校读者我要写书评

暂无评论

Worldwide Influenza Surveillance through Twitter 29

Worldwide Influenza Surveillance through Twitter

引用

29th AAAI Conference on Artificial Intelligence, AAAI 2015

作者： Paul, Michael J. Dredze, Mark Broniatowski, David A. Generous, Nicholas Human Language Technology Center of Excellence Johns Hopkins University BaltimoreMD21218 United States Department of Engineering Management and Systems Engineering George Washington University WashingtonDC20052 United States Los Alamos National Laboratory Defense Systems and Analysis Division Los AlamosNM87545 United States

ISBN: (纸本)9781577357261

We evaluate the performance of Twitter-based influenza surveillance in ten English-speaking countries across four continents. We find that tweets are positively correlated with existing surveillance data provided by government agencies in these countries, with r values ranging from .37-81. We show that incorporating Twitter data into a strong autoregressive baseline reduces mean squared error in 80 to 100 percent of locations depending on the lag, with larger improvements when reporting delays are longer. Copyright © 2015, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词： Mean square error

来源：评论

学校读者我要写书评

暂无评论

Leveraging big data to improve health awareness campaigns: A novel evaluation of the great American smokeout

引用

JMIR Public Health and Surveillance 2016年第1期2卷 e16页

作者： Ayers, John W. Westmaas, J. Lee Leas, Eric C. Benton, Adrian Chen, Yunqi Dredze, Mark Althouse, Benjamin M. Graduate School of Public Health San Diego State University Chula Vista CA United States American Cancer Society Behavioral Research Center Atlanta GA United States University of California San Diego School of Medicine San Diego CA United States Human Language Technology Center of Excellence Johns Hopkins University Baltimore MD United States Bryn Mawr College Bryn Mawr College Philadelphia PA United States Santa Fe Institute Santa Fe NM United States New Mexico State University Las Cruces NM United States

Background: Awareness campaigns are ubiquitous, but little is known about their potential effectiveness because traditional evaluations are often unfeasible. For 40 years, the “Great American Smokeout” (GASO) has encouraged media coverage and popular engagement with smoking cessation on the third Thursday of November as the nation's longest running awareness campaign. Objective: We proposed a novel evaluation framework for assessing awareness campaigns using the GASO as a case study by observing cessation-related news reports and Twitter postings, and cessation-related help seeking via Google, Wikipedia, and government-sponsored quitlines. Methods: Time trends (2009-2014) were analyzed using a quasi-experimental design to isolate spikes during the GASO by comparing observed outcomes on the GASO day with the simulated counterfactual had the GASO not occurred. Results: Cessation-related news typically increased by 61% (95% CI 35-87) and tweets by 13% (95% CI −21 to 48) during the GASO compared with what was expected had the GASO not occurred. Cessation-related Google searches increased by 25% (95% CI 10-40), Wikipedia page visits by 22% (95% CI −26 to 67), and quitline calls by 42% (95% CI 19-64). Cessation-related news media positively coincided with cessation tweets, Internet searches, and Wikipedia visits;for example, a 50% increase in news for any year predicted a 28% (95% CI −2 to 59) increase in tweets for the same year. Increases on the day of the GASO rivaled about two-thirds of a typical New Year's Day-the day that is assumed to see the greatest increases in cessation-related activity. In practical terms, there were about 61,000 more instances of help seeking on Google, Wikipedia, or quitlines on GASO each year than would normally be expected. Conclusions: These findings provide actionable intelligence to improve the GASO and model how to rapidly, cost-effectively, and efficiently evaluate hundreds of awareness campaigns, nearly all for the first time. © John

关键词： Big data Evaluation Health communication Infodemiology Infoveillence Mass media Smoking cessation Social media Tobacco control Twitter

来源：评论

学校读者我要写书评

暂无评论

Quantifying Mental Health Signals in Twitter

Quantifying Mental Health Signals in Twitter

引用

2014 Workshop on Computational Linguistics and Clinical Psychology From Linguistic Signal to Clinical Reality, CLPsych 2014 at the 52nd Annual Meeting of the Association for Computational Linguistics, ACL 2014

作者： Coppersmith, Glen Dredze, Mark Harman, Craig Human Language Technology Center of Excellence Johns Hopkins University BaltimoreMD United States

ISBN: (纸本)9781941643167

The ubiquity of social media provides a rich opportunity to enhance the data available to mental health clinicians and researchers, enabling a better-informed and better-equipped mental health field. We present analysis of mental health phenomena in publicly available Twitter data, demonstrating how rigorous application of simple natural language processing methods can yield insight into specific disorders as well as mental health writ large, along with evidence that as-of-yet undiscovered linguistic signals relevant to mental health exist in social media. We present a novel method for gathering data for a range of mental illnesses quickly and cheaply, then focus on analysis of four in particular: post-traumatic stress disorder (PTSD), depression, bipolar disorder, and seasonal affective disorder (SAD). We intend for these proof-of-concept results to inform the necessary ethical discussion regarding the balance between the utility of such data and the privacy of mental health related information. © 2014 Association for Computational Linguistics

关键词： Social networking (online)

来源：评论

学校读者我要写书评

暂无评论

Measuring post traumatic stress disorder in twitter 8

Measuring post traumatic stress disorder in twitter

引用

8th International Conference on Weblogs and Social Media, ICWSM 2014

作者： Coppersmith, Glen Harman, Craig Dredze, Mark Human Language Technology Center of Excellence Johns Hopkins University BalitmoreMD United States

ISBN: (纸本)9781577356578

Traditional mental health studies rely on information primarily collected through personal contact with a health care professional. Recent work has shown the utility of social media data for studying depression, but there have been limited evaluations of other mental health conditions. We consider post traumatic stress disorder (PTSD), a serious condition that affects millions worldwide, with especially high rates in military veterans. We also present a novel method to obtain a PTSD classifier for social media using simple searches of available Twitter data, a significant reduction in training data cost compared to previous work. We demonstrate its utility by examining differences in language use between PTSD and random individuals, building classifiers to separate these two groups and by detecting elevated rates of PTSD at and around U.S. military bases using our classifiers.. Copyright © 2014, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词： Social networking (online)

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：