Probabilistic symmetries and invariant neural networks

MaRDI QIDQ4969154zbMATH OpenFDO

Authors Benjamin Bloem-Reddy, Yee Whye Teh

Publication date 5 October 2020

Full work available at URL https://arxiv.org/abs/1901.06082, https://jmlr.csail.mit.edu/papers/v21/19-322.html

zbMATH Keywords

exchangeabilityinvariancesufficiencyadequacyconvolutional neural networksequivariancegraph neural networksneural architecturesprobabilistic symmetry

Mathematics Subject Classification ID

Artificial neural networks and deep learning (68T07) Neural nets and related approaches to inference from stochastic processes (62M45) Sufficient statistics and fields (62B05) Exchangeability for stochastic processes (60G09)

Recommendations

• Universal approximations of invariant maps by neural networks
• The complexity of explaining neural networks through (group) invariants
• The complexity of explaining neural networks through (group) invariants
• A unifying framework for invariant pattern recognition
• Representation theory and invariant neural networks

Cites work

• scientific article; zbMATH DE number 1713116 (Why is no real title available?)
• scientific article; zbMATH DE number 6378127 (Why is no real title available?)
• scientific article; zbMATH DE number 3862208 (Why is no real title available?)
• scientific article; zbMATH DE number 3896009 (Why is no real title available?)
• scientific article; zbMATH DE number 46318 (Why is no real title available?)
• scientific article; zbMATH DE number 51418 (Why is no real title available?)
• scientific article; zbMATH DE number 107671 (Why is no real title available?)
• scientific article; zbMATH DE number 3466390 (Why is no real title available?)
• scientific article; zbMATH DE number 3522963 (Why is no real title available?)
• scientific article; zbMATH DE number 1339024 (Why is no real title available?)
• scientific article; zbMATH DE number 706263 (Why is no real title available?)
• scientific article; zbMATH DE number 1134987 (Why is no real title available?)
• scientific article; zbMATH DE number 5021952 (Why is no real title available?)
• A FACTORISATION THEOREM FOR ADEQUATE STATISTICS
• A dozen de Finetti-style results in search of a theory
• Adequate Subfields and Sufficiency
• Application of the Radon-Nikodym Theorem to the Theory of Sufficient Statistics
• Approximation by superpositions of a sigmoidal function
• Bayesian nonparametric analysis of reversible Markov chains
• Bayesian optimization for likelihood-free inference of simulator-based statistical models
• De Finetti's theorem for Markov chains
• Distributions of maximal invariants using quotient measures
• Edge exchangeable models for interaction networks
• Exchangeability, predictive distributions and parametric models
• Exchangeable random measures
• Extremal families and systems of sufficient statistics
• Ferguson distributions via Polya urn schemes
• Finite exchangeable sequences
• Finite forms of de Finetti's theorem on exchangeability
• Group Invariance in Statistical Inference
• Group invariance, stability to deformations, and complexity of deep convolutional representations
• Information Theory and Statistical Mechanics
• Invariance and independence in multivariate distribution theory
• Invariants Under Mixing Which Generalize de Finetti's Theorem: Continuous Time Parameter
• Invariants Under Mixing which Generalize de Finetti's Theorem
• Large networks and graph limits
• Multilayer feedforward networks are universal approximators
• Multiscale sparse microcanonical models
• Nonparametric network models for link prediction
• Practical graph isomorphism. II.
• Probabilistic Symmetries and Invariance Principles
• Probability and stochastics.
• Random measures, theory and applications
• Representation theory and invariant neural networks
• Sample sizes for threshold networks with equivalences
• Sufficiency and Statistical Decision Functions
• Symmetric Measures on Cartesian Products
• Testing Statistical Hypotheses
• The Relationship Between Sufficiency and Invariance with Applications in Sequential Analysis
• Theory of statistics

Cited in

(13)

• On the finite representation of linear group equivariant operators via permutant measures
• Invariance, Encodings, and Generalization: Learning Identity Effects With Neural Networks
• The complexity of explaining neural networks through (group) invariants
• Universal approximations of invariant maps by neural networks
• Iterative SE(3)-transformers
• scientific article; zbMATH DE number 5163730 (Why is no real title available?)
• A de Finetti-type representation of joint hierarchically exchange
• Optimal Dorfman group testing for symmetric distributions
• Group-Invariant Tensor Train Networks for Supervised Learning
• What is... an Equivariant Neural Network?
• Gauge symmetry and neural networks
• The complexity of explaining neural networks through (group) invariants
• Neural networks enforcing physical symmetries in nonlinear dynamical lattices: the case example of the Ablowitz-Ladik model

Describes a project that uses

Uses Software

• Tensor2Tensor
• PyTorch
• TensorFlow
• Theano
• Mathematica
• GAP

This page was built for publication: Probabilistic symmetries and invariant neural networks

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4969154)