Notes:

Markov chain Monte Carlo (MCMC) is a statistical method that is used to sample from a complex probability distribution in order to estimate various quantities of interest. MCMC algorithms work by constructing a Markov chain, which is a sequence of random variables that are related to each other through a set of probabilistic rules. The Markov chain is then used to sample from the target distribution, and the samples are used to estimate various quantities of interest.

In dialog systems, MCMC algorithms can be used to estimate the probability of various dialog outcomes or to make decisions about what actions to take in a given dialog context. For example, an MCMC algorithm might be used to estimate the probability that a given dialog will lead to a successful outcome (e.g., a customer service request being successfully resolved), or to determine the most likely response to a given user input.

MCMC algorithms are particularly useful in dialog systems because they can handle complex, high-dimensional distributions and can estimate quantities of interest even when the underlying distribution is not fully known. This can make them useful for tasks such as language generation, dialogue management, and decision-making in dialog systems.

Wikipedia:

See also:

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