Optimising synaptic learning rules in linear associative memories

DOI10.1007/BF00206223MaRDI QIDQ1182329zbMATH OpenDBLPWikidataFDO

Authors David Willshaw, Peter Dayan

Publication date 28 June 1992

Published in Biological Cybernetics (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1007/BF00206223

zbMATH Keywords

associative matrix memorycovariance rulelearning rulelinear associative memoriesmaximization of the signal-to-noise ratiosynaptic learning rules

Mathematics Subject Classification ID

Learning and adaptive systems in artificial intelligence (68T05) Neural networks for/in biological studies, artificial life and related topics (92B20) Neural biology (92C20)

Review text

A model of associative matrix memory incorporating 4 numbers, defining a learning rule, is considered. The maximization of the signal-to-noise ratio is used as a criterion for optimization of the form of the learning rule of the model. A well known covariance rule is shown to be optimal. Two other rules, previously suggested, are shown to be asymptotically optimal. All three rules, however, are shown to be strongly biologically irrelevant. Some explanation for this discrepancy and further discussion is proposed.

Recommendations

• Neural associative memory with optimal Bayesian learning
• Optimal learning rules for familiarity detection
• Inhomogeneities in Heteroassociative Memories with Linear Learning Rules
• Stability and attractivity in associative memory networks
• Efficient Associative Computation with Discrete Synapses

Cites work

• Correlation Matrix Memories
• scientific article; zbMATH DE number 42973 (Why is no real title available?)
• scientific article; zbMATH DE number 3436645 (Why is no real title available?)
• Neural networks and physical systems with emergent collective computational abilities
• The space of interactions in neural network models

Cited in

(18)

• ``Live neuron and optimal learning rule
• Dendritic morphology predicts pattern recognition performance in multi-compartmental model neurons with and without active conductances
• Studying associative learning without solving learning equations
• Optimal learning rules for familiarity detection
• Learning flexible sensori-motor mappings in a complex network
• A Widrow-Hoff learning rule for a generalization of the linear auto-associator
• Effective neuronal learning with ineffective Hebbian learning rules
• Associative learning using Ising-like model
• Neural associative memory with optimal Bayesian learning
• Memory dynamics in attractor networks with saliency weights
• Binary Willshaw learning yields high synaptic capacity for long-term familiarity memory
• Associative Memory with Dynamic Synapses
• Spike-Timing-Dependent Plasticity and Relevant Mutual Information Maximization
• Iterative retrieval and block coding in autoassociative and heteroassociative memory
• Memory Capacities for Synaptic and Structural Plasticity
• Efficient Associative Computation with Discrete Synapses
• Inhomogeneities in Heteroassociative Memories with Linear Learning Rules
• Hybrid Kolmogorov-Arnold and xLSTM network for enhanced RUL prediction of lithium batteries

This page was built for publication: Optimising synaptic learning rules in linear associative memories

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1182329)