MASSACHUSETTS INSTITUTE OF TECHNOLOGY ARTIFICIAL INTELLIGENCE LABORATORY and CENTER FOR BIOLOGICAL AND COMPUTATIONAL LEARNING DEPARTMENT OF BRAIN AND COGNITIVE SCIENCES

By Zoubin Ghahramani, Michael I. Jordan. 1996

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Composing graphical models with neural networks for structured representations and fast inference (2016, methodology): “[1] [2] [3] [4] Gaussian mixture model Linear dynamical system Hidden Markov model Switching LDS [6] [2] [5] Mixture of Experts Driven LDS IO-HMM Factorial HMM [7]”
Audio-Visual Scene Analysis with Self-Supervised Multisensory Features (2018, background): “Researchers have proposed many successful probabilistic approaches to this problem [30,31,32,33].”
Sequence Classification Restricted Boltzmann Machines With Gated Units (2020, methodology): “The sequence classiﬁcation problem has attracted research in dynamic Bayesian models, such as hidden Markov models (HMMs) [11] and conditional random ﬁelds (CRFs) [12].”
Structured Stochastic Variational Inference (2014, methodology): “Traditional structured mean-field algorithms require the practitioner to identify and exploit some model-specific structure; for example, Ghahramani and Jordan (1997) exploited the availability of dynamic programming algorithms for HMMs to derive a structured mean-field algorithm for factorial…”
An Introduction to Variational Methods for Graphical Models (1999, methodology): “We will not discuss the higher–order HMM further in this paper; for a variational algorithm for the higher–order HMM see Saul and Jordan (1996).”
A Unifying Review of Linear Gaussian Models (1999, methodology): “…are the standard method of illustrating graphical models, also known as probabilistic independence networks, a category of models that includes Markov networks, Bayesian (or belief) networks, and other formalisms (Pearl, 1988; Lauritzen & Spiegelhalter, 1988; Whittaker, 1990; Smyth et al., 1997).”
Conditional models for contextual human motion recognition (2006, methodology): “... recognition structures have been obtained by compositing the basic ones: (i) Layering [22,24] composes recognition modules hierarchically by feeding the results of one recognition layer as observations for the next one; (ii) Left–right modeling allows certain transitions, but not all the possibl”
Learning with Mixtures of Trees (2000, methodology): “We now assign a distribution T to the graph (V, E) as follows: T (x) = (u,v)∈E T uv (x u , x v ) v∈V T v (x v ) deg v−1 , (1) where deg v is the degree of vertex v; i.e., the number of edges incident to v ∈ V .”
Collapsed Variational Bayesian Inference for Hidden Markov Models (2013, background): “Smyth et al. (1997) expressed HMMs as Bayesian networks, which promoted the development of a number of Bayesian approaches (MacKay, 1997; Beal, 2003; Goldwater and Griffiths, 2007).”
Dialog state tracking, a machine reading approach using Memory Network (2016, methodology): “A typical generative model uses a factorial hidden Markov model [6].”
The Infinite Factorial Hidden Markov Model (2008, background): “Another issue which deserves further exploration is inference: in [2] it was found that a structured variational method provides a good balance between accuracy and computational effort.”
Dynamic Graphical Models (2010, background): “The following techniques generalize those mentioned in the section “HMMs and Graphical Models” and consist of two phases.”
An Auxiliary Variational Method (2004, methodology): “One way to go beyond the factorized assumption for q ( x ) is to consider a structured mean ﬂeld approximation (Ghahramani and Jordan, 1995; Barber and Wiegerinck, 1998) which retains some of the structure of p ( x ).”
Markovian Models for Sequential Data (2004, methodology): “See also [46, 47] for relations between HMMs and other graphical models such as Markov random elds and Kalman lters, and see [48] for an application of these ideas to Turbo-decoding.”