Recognition Networks for Approximate Inference in BN20 Networks

A recognition network is a multilayer perception (MLP) trained to predict posterior maxginals given observed evidence in a particulax Bayesian network. The input to the MLP is a vector of the states of the evidential nodes. The activity of an output unit is interpreted as a prediction of the posterior marginal of the corresponding variable. The MLP is trained using samples generated from the corresponding Bayesian network. We evaluate a recognition network that was trained to do inference in a large Bayesian network, similax in structure and complexity to the Quick Medical Reference, Decision Theoretic (QMR-DT) network. Our network is a binary, two-layer, noisy-OR (BN20) network containing over 4000 potentially observable nodes and over 600 unobservable, hidden nodes. In real medical diagnosis, most observables are unavailable, and there is a complex and unknown process that selects which ones axe provided. We incorporate a very basic type of selection bias in our network: a known preference that available observables are positive rather than negative. Even this simple bias has a significant effect on the posterior. We compare the performance of our recognition network to state-of-the-art approximate inference algorithms on a large set of test cases. In order to evaluate the effect of our simplistic model of the selection bias, we evaluate algorithms using a variety of incorrectly modelled selection biases. Recognition networks perform well using both correct and incorrect selection biases.

• Publication year

  2001

• Venue

  Conference on Uncertainty in Artificial Intelligence

• Publication date

  2001-08-02

• Fields of study

  Computer Science

• Identifiers
  arXiv 1301.2295
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• A Tractable Inference Algorithm for Diagnosing Multiple Diseases

  2013cited by this paper
• Variational Probabilistic Inference and the QMR-DT Network

  2011cited by this paper
• Conformant Planning via Symbolic Model Checking

  2000cited by this paper
• Sequentially Fitting "Inclusive" Trees for Inference in Noisy-OR Networks

  2000cited by this paper
• Loopy Belief Propagation for Approximate Inference: An Empirical Study

  1999cited by this paper
• Approximate Inference A lgorithms for Two-Layer Bayesian Networks

  1999cited by this paper
• Centering Neural Network Gradient Factors

  1996cited by this paper
• The "wake-sleep" algorithm for unsupervised neural networks.

  1995cited by this paper
• The Helmholtz Machine

  1995cited by this paper
• Symbolic Probabilistic Inference in Large BN20 Networks

  1994cited by this paper
• An Empirical Analysis of Likelihood-Weighting Simulation on a Large, Multiply-Connected Belief Network

  1991cited by this paper
• Probabilistic diagnosis using a reformulation of the INTERNIST-1/QMR knowledge base. II. Evaluation of diagnostic performance.

  1991cited by this paper
• Probabilistic Diagnosis Using a Reformulation of the INTERNIST-1/QMR Knowledge Base

  1991cited by this paper
• How Neural Nets Work

  1987cited by this paper
• Internist-I, an Experimental Computer-Based Diagnostic Consultant for General Internal Medicine

  1982cited by this paper

• Amortized In-Context Bayesian Posterior Estimation

  2025cites this paper
• Evaluating the Diagnostic Performance of Symptom Checkers: Clinical Vignette Study

  2023cites this paper
• Intrusion detection framework based on causal reasoning for DDoS

  2022cites this paper
• Author Correction: Improving the accuracy of medical diagnosis with causal machine learning

  2021cites this paper
• Variational inference with a quantum computer

  2021cites this paper
• COVID-19 in differential diagnosis of online symptom assessments

  2020cites this paper
• Improving the accuracy of medical diagnosis with causal machine learning

  2020cites this paper
• Reconstructing the Evolutionary History of Tumours

  2019cites this paper
• Amortized Inference of Variational Bounds for Learning Noisy-OR

  2019cites this paper
• Counterfactual diagnosis

  2019cites this paper
• Universal Marginaliser for Deep Amortised Inference for Probabilistic Programs

  2019cites this paper
• MultiVerse: Causal Reasoning using Importance Sampling in Probabilistic Programming

  2019cites this paper
• Neural Density Estimation and Likelihood-free Inference

  2019cites this paper
• The accuracy vs. coverage trade-off in patient-facing diagnosis models

  2019cites this paper
• Sequential Neural Likelihood: Fast Likelihood-free Inference with Autoregressive Flows

  2018cites this paper
• Approximate inference: new visions

  2018cites this paper
• Universal Marginalizer for Amortised Inference and Embedding of Generative Models

  2018cites this paper
• Learning from the experts: From expert systems to machine learned diagnosis models

  2018cites this paper
• Probabilistic programs for inferring the goals of autonomous agents

  2017cites this paper
• A Universal Marginalizer for Amortized Inference in Generative Models

  2017cites this paper
• Encapsulating models and approximate inference programs in probabilistic modules

  2016cites this paper
• Fast $\epsilon$-free Inference of Simulation Models with Bayesian Conditional Density Estimation

  2016cites this paper
• Modeling Cognition with Probabilistic Programs: Representations and Algorithms

  2015cites this paper
• Learning Stochastic Inverses

  2013cites this paper
• Fast Variational Inference for Large-scale Internet Diagnosis

  2007cites this paper
• Bethe free energy and contrastive divergence approximations for undirected graphical models

  2003influential citation
• Observation Biases in Diagnostic Inference

  2001cites this paper
• Exploring Exchangeable Dataset Amortization for Bayesian Posterior Inference

  year unknowncites this paper