MCMC (Markov Chain Monte Carlo) & Dialog Systems


MCMC (Markov Chain Monte Carlo) & Dialog Systems
Markov chain Monte Carlo 


Reinforcement learning with limited reinforcement: Using Bayes risk for active learning in POMDPs F Doshi-Velez, J Pineau… – Artificial Intelligence, 2012 – Elsevier … The agentâ€(tm)s actions may also have unexpected results: in the case of the dialog system, aquestion from the system … We use a Markov chain Monte Carlo approach, forward-filtering backward-sampling (FFBS) [31] to sample dynamics models from this proposal distribution K …

Towards relational POMDPs for adaptive dialogue management [PDF] from aclweb.org P Lison – Proceedings of the ACL 2010 Student Research …, 2010 – dl.acm.org … Furthemore, the dialogue system also needs to be adaptive to its user (at- tributed beliefs and intentions, attitude, attentional state … Efficient probabilistic inference algorithms such as Markov Chain Monte Carlo (MCMC) or other sampling techniques can then be used to this end … Cited by 1 – Related articles – All 21 versions

Are You Sure You’re Paying Attention?’-Uh-Huh’Communicating Understanding as a Marker of Attentiveness [PDF] from uni-bielefeld.de H Buschmeier, Z Malisz, M Wlodarczak… – … Annual Conference of …, 2011 – isca-speech.org … We argue this finding could be used to facilitate recognition of attentional states in dialogue system users. … We report parameter estimates in Table 2. p- values were calculated by means of Markov Chain Monte Carlo (MCMC) sampling. … Cited by 1 – Related articles – All 5 versions

Call for Participation CE Rasmussen… – 2010 – Springer … D. Jain, M. Beetz: Soft Evidential Update via Markov Chain Monte Carlo Inference … Several approaches have been made con- cerning emotion recognition, emotion modelling, genera- tion of emotional user interfaces and dialogue systems. …

AI’s 10 to Watch [PDF] from stanford.edu J Hendler, P Cimiano, D Dolgov… – Intelligent Systems, …, 2008 – ieeexplore.ieee.org … Such an ontol- ogy-based approach to language processing has important applications not only in ques- tion answering, information extraction, and dialogue systems but also in the Semantic Web, which requires that today’s Web con- tent is formalized using ontologies. … Cited by 1 – Related articles – All 16 versions

A two-dimensional topic-aspect model for discovering multi-faceted topics [PDF] from uiuc.edu M Paul… – Urbana, 2010 – aaai.org … is intractable. We will instead approximate this using Gibbs sampling, a Markov chain Monte Carlo algo- rithm. In a … neutral distribution. The CL aspect focuses on dialogue systems, with words like dialogue and user. The LING … Cited by 13 – Related articles – All 6 versions

What Do You Mean, You’re Uncertain?: The Interpretation of Cue Words and Rising Intonation in Dialogue [PDF] from upenn.edu C Lai – Eleventh Annual Conference of the International …, 2010 – isca-speech.org … model parameters, along with finite popu- lation standard deviations for each group, were estimated using the Markov Chain Monte Carlo technique as … compose has implications for both formal the- ories of dialogue and for determining how a dialogue system should respond to … Cited by 1 – Related articles – All 6 versions

Grammatical error simulation for computer-assisted language learning S Lee, J Lee, H Noh, K Lee… – Knowledge-Based Systems, 2011 – Elsevier … Recently, user simulation has become widely used in the development of spoken dialog systems to develop dialog strategies … In principle, probabilistic inference can be performed using belief propagation, Markov Chain Monte Carlo (MCMC), the variational approximation, and … Cited by 3 – Related articles – All 3 versions

[CITATION] Deliverable 4.2 S Pulman, R Granell… – 2008 Related articles – All 10 versions

Towards Link Characterization From Content: Recovering Distributions From Classifier Output J Grothendieck… – Audio, Speech, and Language …, 2008 – ieeexplore.ieee.org … techniques to fit a particular mixture problem. The justification for using Markov Chain Monte Carlo (MCMC) numerical estimation is well-understood [6], but the practice involves some art [37]-[39]. For a classifier that gave accurate … Cited by 1 – Related articles – BL Direct – All 4 versions

A probabilistic inference of multiparty-conversation structure based on Markov-switching models of gaze patterns, head directions, and utterances [PDF] from ntt.co.jp K Otsuka, Y Takemae… – … of the 7th international conference on …, 2005 – dl.acm.org … General Terms: ALGORITHMS, HUMAN FACTORS Keywords: Face-to-face multiparty conversation, Eye gaze, Nonverbal cues, Dynamic Bayesian network, Markov-switching model, Markov chain Monte Carlo, Gibbs sampler … Cited by 50 – Related articles – All 6 versions

[PDF] Belief modelling for situation awareness in human-robot interaction [PDF] from uio.no P Lison, C Ehrler… – … of the 19th International Symposium on …, 2010 – folk.uio.no … untractable. However, several efficient algorithms for probabilistic in- ference such as weighted MAX-SAT, Markov Chain Monte Carlo (MCMC) or lifted belief propagation can then be used to yield approximate solutions [23,25,29]. … Cited by 7 – Related articles – View as HTML – All 11 versions

[PDF] Cross-collection topic models: Automatically comparing and contrasting text [PDF] from jhu.edu M Paul – Urbana, 2009 – cs.jhu.edu … Gibbs sampling is a type of Markov chain Monte Carlo algorithm and is what we employ in this paper, as it is simple to derive, comparable in … We see that in CL, this is strongly relevant to dialogue systems; in linguistics, this topic is more focused on human behavior and social … Cited by 2 – Related articles – View as HTML – All 2 versions

[PDF] Dialogue with attitude: The contribution of cue words and prosodic meaning in conversational speech [PDF] from upenn.edu C Lai – 2011 – ling.upenn.edu Page 1. Dialogue with attitude: The contribution of cue words and prosodic meaning in conversational speech Catherine Lai Department of Linguistics University of Pennsylvania April 2011 Lai (University of Pennsylvania) Cue Words and Prosody 1 / 64 Page 2. … Related articles – View as HTML

[PDF] Offering online recommendations with minimum customer input through conjoint-based decision aids [PDF] from garylilien.info A De Bruyn, JC Liechty, EKRE Huizingh… – Marketing Science, 2008 – garylilien.info … In contrast to traditional tree methods that apply locally optimal, greedy splitting rules, Bayesian treed regression tries to achieve global optimality by search- ing the space of possible trees using Markov chain Monte Carlo (MCMC) exploration (Chipman et al. 2002). … Cited by 25 – Related articles – View as HTML – All 13 versions

[PDF] Discriminative Adaptive Training and Bayesian Inference for Speech Recognition [PDF] from psu.edu CK Raut – Emmanuel College, University of Cambridge, 2009 – Citeseer … 39 2.6.3 Bayesian Inference . . . . . 41 2.6.3.1 Markov Chain Monte-Carlo . . . . . 41 2.6.3.2 Frame-Independence Assumption . . . . . 42 2.6.3.3 Laplace Approximation . . . . . 42 … Cited by 1 – Related articles – View as HTML – All 6 versions

Unsupervised Segmentation of Human Motion Data Using a Sticky Hierarchical Dirichlet Process-Hidden Markov Model and Minimal Description Length-Based … [PDF] from nict.go.jp T Taniguchi, K Hamahata… – Advanced Robotics, 2011 – ingentaconnect.com … complexity and flexi- bility. Therefore, a learning method that guarantees global optimality is expected to be developed. The Markov chain Monte-Carlo (MCMC) algorithm was introduced to overcome this problem [27]. In addition, it is … Related articles – All 2 versions

Probabilistic inference of gaze patterns and structure of multiparty conversations from head directions and utterances [PDF] from nagoya-u.ac.jp K Otsuka, Y Takemae, J Yamato… – New Frontiers in Artificial …, 2006 – Springer … The regime changes over time are based on Markov transitions, and controls the dynamics of the gaze pat- terns and utterances. Furthermore, Bayesian estimation of regime, gaze pattern, and model parameters are implemented using a Markov chain Monte Carlo method. … Cited by 2 – Related articles – BL Direct – All 15 versions

Context effects in language production: Models of syntactic priming in dialogue corpora [PDF] from ed.ac.uk D Reitter – 2008 – era.lib.ed.ac.uk Page 1. Context Effects in Language Production: Models of Syntactic Priming in Dialogue Corpora David Reitter School of Informatics, University of Edinburgh T H E U NIVER S I T Y O F E DI NBU R G H Doctor of Philosophy … Cited by 17 – Related articles – All 6 versions

[PDF] Human Language Technology Conference of the North American Chapter of the Association of Computational Linguistics [PDF] from aclweb.org C Manning, D Oard… – 2006 – aclweb.org … is a researcher in discourse/dialogue, entity extraction, and evaluation of machine translation and dialogue systems. … Integration – Integration by Summing – Monte Carlo Integration 5. Advanced Bayesian Inference Techniques – Markov Chain Monte Carlo Integration – Laplace … View as HTML – All 19 versions

An error-corrective language-model adaptation for automatic speech recognition [PDF] from minwoojeong.com M Jeong, J Eun, S Jung… – Ninth European Conference on …, 2005 – isca-speech.org … Sampling Method Several Markov chain monte carlo (MCMC) sampling tech- niques have been investigated for WSME-LM training [7]. Im- portance … made by University of Colorado is an over-the-telephone spoken dialogue corpus to develop a dialogue system for accessing … Cited by 5 – Related articles – All 7 versions

Analysis of Individual Risk Attitude for Risk Management Based on Cumulative Prospect Theory FC Hsu… – … Pattern Discovery and Recovery, New York, …, 2009 – books.google.com Page 277. Chapter XV Analysis of Individual Risk Attitude for Risk Management Based on Cumulative Prospect Theory Fei-Chen Hsu National Tsing Hua University, Taiwan, ROC Hsiao-Fan Wang National Tsing Hua University … Cited by 2 – Related articles – All 6 versions

Empathic Touch by Relational Agents [HTML] from computer.org TW Bickmore, R Fernando, L Ring… – … , IEEE Transactions on, 2010 – ieeexplore.ieee.org … squeezing, stroking, etc.) by the patient and able to use these same communicative signals in conjunction with a speech-based dialogue system for comforting … Significance tests are derived from Markov Chain Monte Carlo sampling, performed with the languageR [44] package. … Cited by 4 – Related articles – All 16 versions

Improving Bayesian network structure learning with mutual information-based node ordering in the K2 algorithm XW Chen, G Anantha… – Knowledge and Data …, 2008 – ieeexplore.ieee.org … these variables [1]. It has been applied to a wide range of tasks such as natural spoken dialog systems [2], vision … These include heuristic search techniques [12], [23], [34], [35], genetic algorithms [36], [37], simulated annealing [38], and Markov Chain Monte Carlo (MCMC) [25 … Cited by 34 – Related articles – BL Direct – All 7 versions

[PDF] Bayesian Recommender Systems: Models and Algorithms [PDF] from anu.edu.au S Guo – 2011 – users.cecs.anu.edu.au Page 1. Bayesian Recommender Systems: Models and Algorithms Shengbo Guo October 2011 A thesis submitted for the degree of Doctor of Philosophy of The Australian National University Page 2. Page 3. Declaration These … Related articles – View as HTML

[PDF] Prior Knowledge for Text and Language Processing [PDF] from helsinki.fi G Bouchard, H Daumé III, M Dymetman… – 2008 – uai2008.cs.helsinki.fi Page 1. Proceedings of the Workshop on Prior Knowledge for Text and Language Processing Organizers: Guillaume Bouchard Hal Daumé III Marc Dymetman Yee Whye Teh An ICML/UAI/COLT Workshop Helsinki, Finland 9 July 2008 Page 2. Organizers: … Related articles – View as HTML – All 9 versions

Bayesian Networks for Discrete Observation Distributions in Speech Recognition A Miguel, A Ortega, L Buera… – Audio, Speech, and …, 2011 – ieeexplore.ieee.org … of this family of techniques in the speech community, good performances have been obtained when used as an alternative to the HMM independence assumption, [12], to model the covariance ma- trix structure [13], [14], for language modeling [15] or in dialog systems [16]. … Cited by 2 – Related articles – All 2 versions

Language Modeling for limited-data domains [PDF] from mit.edu JR Glass, BJ Hsu – 2009 – dspace.mit.edu … In a collaborative environment like Wikipedia, over half of the edits are made by less than 0.7% of the registered users [147]. Similarly, we also expect a small fraction of enthusiasts of such a feedback-driven spoken dialog system to contribute the majority of the training data. … Related articles – All 3 versions

Learning, logic, and probability: A unified view P Domingos – Lecture Notes in Computer Science, 2004 – books.google.com … Inference is performed by Markov chain Monte Carlo over the minimal subset of the ground network required for answering the query. … 2004. Vol. 3068: E. Andre, L. Dybkjser, W. Minker, P. Heis- terkamp (Eds.), Affective Dialogue Systems. XII, 324 pages. 2004. Vol. … Cited by 1 – Related articles – BL Direct – All 4 versions

[PDF] Language Modeling for Limited-Data Domains [PDF] from mit.edu BJP Hsu – 2009 – groups.csail.mit.edu … In a collaborative environment like Wikipedia, over half of the edits are made by less than 0.7% of the registered users [147]. Similarly, we also expect a small fraction of enthusiasts of such a feedback-driven spoken dialog system to contribute the majority of the training data. … Cited by 1 – Related articles – View as HTML – Library Search – All 2 versions

A BAYESIAN TUTORING SYSTEM FOR NEWTONIAN MECHANICS: CAN IT ADAPT TO DIFFERENT LEARNERS? PK Pek, KL Poh – Journal of Educational Computing Research, 2004 – Baywood … Using past data of students’ performance in formal examinations, the person and item parameters can be calibrated through standard statistical technique such as Markov chain Monte Carlo simulation (Patz & Junker, 1997). … Cited by 5 – Related articles – All 3 versions

[PDF] Ph. D Year 1: Transfer Report (RD5) Towards a Patient Simulator Framework for Evaluation of e-Health Environments: Modeling, Techniques, Validation and … [PDF] from networksims.com O Lo – 2011 – networksims.com Page 1. Ph.D Year 1: Transfer Report (RD5) Towards a Patient Simulator Framework for Evaluation of e-Health Environments: Modeling, Techniques, Validation and Implementation Owen Lo February 2011 Page 2. Contents 1 Introduction 1 1.1 Introduction . . . . . … Related articles – View as HTML – All 9 versions

Analysis of human behavior to a communication robot in an open field S Nabe, T Kanda, K Hiraki, H Ishiguro… – Proceedings of the 1st …, 2006 – dl.acm.org … 527-532, 2005 [8] Fujie, S., Ejiri, Y., Matsusaka, Y., Kikuchi, H., Kobayashi, T., Recognition of Para-linguistic Information and its Application to Spoken Dialogue System, IEEE Workshop on Automatic Speech Recognition and Understanding (ASRU ’03), pp. … Cited by 9 – Related articles – All 2 versions

[BOOK] Medinfo 2007 KA Kuhn – 2007 – books.google.com … AdaRTE: Adaptable Dialogue Architecture and Runtime Engine. A New Architecture for Health-Care Dialogue Systems 1063 LM Rojas-Barahona and T. Giorgino Page 18. Multi-Channel Physiological Sensing of Human Emotion … Cited by 4 – Related articles – All 2 versions

Saying what’s on your mind: Working memory effects on sentence production. [PDF] from umd.edu LR Slevc – 2011 – psycnet.apa.org Page 1. Saying What’s on Your Mind: Working Memory Effects on Sentence Production L. Robert Slevc University of Maryland, College Park The role of working memory (WM) in sentence comprehension has received considerable … Cited by 1 – Related articles – All 7 versions

CoCITe-Coordinating Changes in Text [HTML] from computer.org J Wright… – Knowledge and Data Engineering, …, 2012 – ieeexplore.ieee.org Page 1. CoCITe-Coordinating Changes in Text Jeremy H. Wright, Member, IEEE, and John Grothendieck, Member, IEEE Abstract-Text streams are ubiquitous and contain a wealth of information, but are typically orders of … Related articles – All 18 versions

[PDF] LEARNING SUB-WORD UNITS AND EXPLOITING CONTEXTUAL INFORMATION FOR OPEN VOCABULARY SPEECH RECOGNITION [PDF] from jhu.edu MC Parada – 2011 – old-site.clsp.jhu.edu Page 1. LEARNING SUB-WORD UNITS AND EXPLOITING CONTEXTUAL INFORMATION FOR OPEN VOCABULARY SPEECH RECOGNITION by Maria Carolina Parada A dissertation submitted to The Johns Hopkins University in conformity with the … Related articles – View as HTML – All 2 versions

Challenges for discrete mathematics and theoretical computer science in the defense against bioterrorism [PS] from rutgers.edu FS Roberts – … and Modeling Approaches in Homeland Security, …, 2003 – books.google.com … Markov chain Monte Carlo methods are promising tools used to fit data to models of numerous diseases (see, eg,[100, 178]). Moment closure methods have been useful in modeling sexually transmitted diseases, tuberculosis, and other diseases (see, eg,[127]). … Cited by 10 – Related articles – All 4 versions

[PDF] Affective state detection with dynamic bayesian networks [PDF] from tudelft.nl MA de Jongh – 2005 – kbs.twi.tudelft.nl Page 1. ing. MA de Jongh 8 December 2005 Man-Machine Interaction Group Media and Knowledge Engineering Faculty of Electrical Engineering, Mathematics and Computer Science Delft University of Technology Affective State Detection With Dynamic Bayesian Networks … Cited by 1 – Related articles – View as HTML

[PDF] MediaHub: An Intelligent MultiMedia Distributed Platform Hub [PDF] from paulmckevitt.com GG Campbell, T Lunney… – 2005 – paulmckevitt.com … model. 2.3.9 SmartKom SmartKom (Wahlster 2003, Wahlster et al. 2001, SmartKom 2005) is a multimodal dialogue system that is being developed to help overcome the problems of interaction between people and machines. The … Related articles – View as HTML – All 8 versions

[BOOK] Bioterrorism: mathematical modeling applications in homeland security HT Banks… – 2003 – books.google.com … Markov chain Monte Carlo methods are promising tools used to fit data to models of numerous diseases (see, eg, [100, 178]). Moment closure methods have been useful in modeling sexually transmitted diseases, tuberculosis, and other diseases (see, eg, [127]). … Cited by 17 – Related articles – Library Search – All 4 versions

[PDF] Saying what’s on your mind: Working memory effects on syntactic production [PDF] from escholarship.org LR Slevc – 2007 – escholarship.org Page 1. … Cited by 6 – Related articles – View as HTML – All 4 versions

[PDF] RIACS FY2002 Annual Report [PDF] from riacs.edu BM Leiner – 2002 – riacs.edu Page 1. RIACS FY2002 Annual Report Barry M. Leiner RIACS Technical Report AR-02 November 2002 Page 2. RIACS FY2002 Annual Report October 2001 through September 2002 THIS PAGE IS INTENTIONALLY LEFT BLANK Page 3. … Related articles – View as HTML – All 3 versions

A framework for exploiting electronic documentation in support of innovation processes [TXT] from sun.ac.za JW Uys – 2010 – scholar.sun.ac.za Stellenbosch University Department of Industrial Engineering A Framework for Exploiting Electronic Documentation in Support of Innovation Processes JW Uys Dissertation presented for the degree of Doctor of Philosophy at Stellenbosch University. Promoter: Prof. … Cited by 3 – Related articles – All 6 versions