3,170 Hits in 2.7 sec

Tractable Bayesian learning of tree belief networks

Marina Meilă, Tommi Jaakkola
2006 Statistics and computing  
In this paper we present decomposable priors, a family of priors over structure and parameters of tree belief nets for which B a yesian learning with complete observations is tractable, in the sense that  ...  Besides allowing for exact Bayesian learning, these results permit us to formulate a new class of tractable latent v ariable models in which the likelihood of a data point is computed through an ensemble  ...  Discussion This paper has presented decomposable priors, a class of priors over tree structures and parameters that makes exact Bayesian learning tractable.  ... 
doi:10.1007/s11222-006-5535-3 fatcat:6l7ai7ot4bfffaz3vmedzadlai

The Libra Toolkit for Probabilistic Models [article]

Daniel Lowd, Amirmohammad Rooshenas
2015 arXiv   pre-print
The Libra Toolkit is a collection of algorithms for learning and inference with discrete probabilistic models, including Bayesian networks, Markov networks, dependency networks, and sum-product networks  ...  Compared to other toolkits, Libra places a greater emphasis on learning the structure of tractable models in which exact inference is efficient.  ...  The views and conclusions contained in this document are those of the authors and should not be interpreted as necessarily representing the official policies, either expressed or implied, of ARO, NIH,  ... 
arXiv:1504.00110v1 fatcat:ez76sv6mergybbra4ecjdbeupy

Look Ma, No Latent Variables: Accurate Cutset Networks via Compilation

Tahrima Rahman, Shasha Jin, Vibhav Gogate
2019 International Conference on Machine Learning  
Our approach addresses a major limitation of existing techniques that learn cutset networks from data in that their accuracy is quite low as compared to latent variable models such as ensembles of cutset  ...  The key idea in our approach is to construct deep cutset networks by not only learning them from data but also compiling them from a more accurate latent tractable model.  ...  Any opinions, findings, conclusions or recommendations expressed in this paper are those of the authors and do not necessarily reflect the views or official policies, either expressed or implied, of DARPA  ... 
dblp:conf/icml/RahmanJG19 fatcat:fy3nh5ntgrgmhbgljoopdsqu54

Cutset Bayesian Networks: A New Representation for Learning Rao-Blackwellised Graphical Models

Tahrima Rahman, Shasha Jin, Vibhav Gogate
2019 Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence  
The main idea in CBNs is to partition the variables into two subsets X and Y, learn a (intractable) Bayesian network that represents P(X) and a tractable conditional model that represents P(Y|X).  ...  In this paper, we seek to further explore this trade-off between generalization performance and inference accuracy by proposing a novel, partially tractable representation called cutset Bayesian networks  ...  Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence  ... 
doi:10.24963/ijcai.2019/797 dblp:conf/ijcai/RahmanJG19 fatcat:fpxg776ig5g3rptbvmrd32ebaa

Cutset Networks: A Simple, Tractable, and Scalable Approach for Improving the Accuracy of Chow-Liu Trees [chapter]

Tahrima Rahman, Prasanna Kothalkar, Vibhav Gogate
2014 Lecture Notes in Computer Science  
Cutset networks are rooted OR search trees, in which each OR node represents conditioning of a variable in the model, with tree Bayesian networks (Chow-Liu trees) at the leaves.  ...  Our experiments on a wide variety of benchmark datasets clearly demonstrate that compared to approaches for learning other tractable models such as thinjunction trees, latent tree models, arithmetic circuits  ...  The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies, either expressed or implied, of DARPA, AFRL, ARO or the  ... 
doi:10.1007/978-3-662-44851-9_40 fatcat:lfll64st2ncp7lhu3nq3at7vd4

Propagation Algorithms for Variational Bayesian Learning

Zoubin Ghahramani, Matthew J. Beal
2000 Neural Information Processing Systems  
We show how the belief propagation and the junction tree algorithms can be used in the inference step of variational Bayesian learning.  ...  Variational approximations are becoming a widespread tool for Bayesian learning of graphical models.  ...  For conjugate-exponential models, integrating both belief propagation and the junction tree algorithm into the variational Bayesian framework simply amounts to computing expectations of the natural parameters  ... 
dblp:conf/nips/GhahramaniB00 fatcat:bhmon227hndbfhywajhlt7ktvq

Learning without recall in directed circles and rooted trees

Mohammad Amin Rahimian, Ali Jadbabaie
2015 2015 American Control Conference (ACC)  
This way, one can realize an exponentially fast rate of learning similar to the case of Bayesian (fully rational) agents. The proposed rules are a special case of the Learning without Recall.  ...  This work investigates the case of a network of agents that attempt to learn some unknown state of the world amongst the finitely many possibilities.  ...  The authors' ongoing research focuses on the investigation and analysis of belief update rules that provide asymptotic learning in a wider variety of network structures and facilitate the tractable modeling  ... 
doi:10.1109/acc.2015.7171992 dblp:conf/amcc/RahimianJ15 fatcat:xi6e47gf2rf4zhecg3bo3g375e

A Non-parametric Bayesian Network Prior of Human Pose

Andreas M. Lehrmann, Peter V. Gehler, Sebastian Nowozin
2013 2013 IEEE International Conference on Computer Vision  
In this work, we introduce a sparse Bayesian network model of human pose that is non-parametric with respect to the estimation of both its graph structure and its local distributions.  ...  We describe an efficient sampling scheme for our model and show its tractability for the computation of exact log-likelihoods.  ...  Non-parametric Bayesian Networks In this section we introduce our non-parametric Bayesian network model of human pose and show its tractability.  ... 
doi:10.1109/iccv.2013.162 dblp:conf/iccv/LehrmannGN13 fatcat:cmtsktin3vgbtnevvpijc4akwe

Approximate Inference by Compilation to Arithmetic Circuits

Daniel Lowd, Pedro M. Domingos
2010 Neural Information Processing Systems  
We propose and evaluate a variety of techniques based on exact compilation, forward sampling, AC structure learning, Markov network parameter learning, variational inference, and Gibbs sampling.  ...  In experiments on eight challenging real-world domains, we find that the methods based on sampling and learning work best: one such method (AC 2 -F) is faster and usually more accurate than loopy belief  ...  The views and conclusions contained in this document are those of the authors and should not be interpreted as necessarily representing the official policies, either expressed or implied, of ARO, DARPA  ... 
dblp:conf/nips/LowdD10 fatcat:3vugsjcgrnbndi6l7rftmnidwe

Bayesian update of dialogue state for robust dialogue systems

Blaise Thomson, Jost Schatzmann, Steve Young
2008 Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing  
The technique is based on updating a Bayesian Network that represents the underlying state of a Partially Observable Markov Decision Process (POMDP).  ...  This paper presents a new framework for accumulating beliefs in spoken dialogue systems.  ...  Figure 1 shows an example network for two time-slices of a two-slot system based on this idea. Belief updating is done with standard Dynamic Bayesian Network algorithms.  ... 
doi:10.1109/icassp.2008.4518765 dblp:conf/icassp/ThomsonSY08 fatcat:okszuy4ynzgrfo4obtcpa7xcei

Recent Advances in Probabilistic Graphical Models

Concha Bielza, Serafín Moral, Antonio Salmerón
2015 International Journal of Intelligent Systems  
Bayesian networks are the most prominent type of probabilistic graphical models and have experienced a remarkable methodological development during the past two decades.  ...  Regardless of the increasing interest in the area, probabilistic graphical models are still facing a number of challenges, covering modeling, inference and learning.  ...  Herna´ndez-Gonza´lez et al. learn the parameters of a multidimensional Bayesian network classifier when the data are (subjectively) labeled by a crowd of annotators.  ... 
doi:10.1002/int.21697 fatcat:cvmcvjrzqvhyfa2u6hxixrqyhq

Naive Bayes models for probability estimation

Daniel Lowd, Pedro Domingos
2005 Proceedings of the 22nd international conference on Machine learning - ICML '05  
In this paper we show that, for a wide range of benchmark datasets, naive Bayes models learned using EM have accuracy and learning time comparable to Bayesian networks with context-specific independence  ...  However, they are seldom used for general probabilistic learning and inference (i.e., for estimating and computing arbitrary joint, conditional and marginal distributions).  ...  Belief propagation is a message-passing algorithm originally used to perform exact inference on tree-structured Bayesian networks.  ... 
doi:10.1145/1102351.1102418 dblp:conf/icml/LowdD05 fatcat:tygpblqf4fg5zhmk24rozmcwpy

Dynamic Data Feed to Bayesian Network Model and SMILE Web Application

Nipat Jongsawat, Pittaya Poompuang, Wichian Premchaiswadi
2008 2008 Ninth ACIS International Conference on Software Engineering, Artificial Intelligence, Networking, and Parallel/Distributed Computing  
They are the followings: Bayesian Network Bayesian networks (also called belief networks, Bayesian belief networks, causal probabilistic networks, or causal networks) (Pearl, 1988 ) are acyclic directed  ...  There exist several efficient algorithms, however, that make belief updating in graphs consisting of tens or hundreds of variables tractable.  ...  Bayesian networks are a very general and powerful tool that can be used for a large number of problems involving uncertainty: reasoning, learning, planning and perception.  ... 
doi:10.1109/snpd.2008.67 dblp:conf/snpd/JongsawatPP08 fatcat:p3ei2ffpxrcrxjhrbwn2ovxtau

Approximating Posterior Distributions in Belief Networks Using Mixtures

Christopher M. Bishop, Neil D. Lawrence, Tommi S. Jaakkola, Michael I. Jordan
1997 Neural Information Processing Systems  
We derive an efficient algorithm for updating the mixture parameters and apply it to the problem of learning in sigmoid belief networks.  ...  Exact inference in densely connected Bayesian networks is computationally intractable, and so there is considerable interest in developing effective approximation schemes.  ...  Introduction Bayesian belief networks can be regarded as a fully probabilistic interpretation of feedforward neural networks.  ... 
dblp:conf/nips/BishopLJJ97 fatcat:fbhwx3kmdrgitnv3r73fchrctm

Locally Bayesian learning in networks

Wei Li, Xu Tan
2020 Theoretical Economics  
We present a tractable learning rule to implement such locally Bayesian learning: each agent extracts new information using the full history of observed reports in her local network.  ...  Despite their limited network knowledge, agents learn correctly when the network is a social quilt, a tree‐like union of cliques.  ...  Moreover, locally Bayesian learning is far more tractable than Bayesian learning and is thus potentially useful for other network learning models. Our model can be extended in several directions.  ... 
doi:10.3982/te3273 fatcat:xdjtnd2gx5davd3vv3dwcwsoqm
« Previous Showing results 1 — 15 out of 3,170 results