Litcius/Paper detail

Analysis and Design of Multivalued High-Capacity Associative Memories Based on Delayed Recurrent Neural Networks

Jiahui Zhang, Song Zhu, Gang Bao, Xiaoyang Liu, Shiping Wen

2021IEEE Transactions on Cybernetics26 citationsDOI

Abstract

This article aims at analyzing and designing the multivalued high-capacity-associative memories based on recurrent neural networks with both asynchronous and distributed delays. In order to increase storage capacities, multivalued activation functions are introduced into associative memories. The stored patterns are retrieved by external input vectors instead of initial conditions, which can guarantee accurate associative memories by avoiding spurious equilibrium points. Some sufficient conditions are proposed to ensure the existence, uniqueness, and global exponential stability of the equilibrium point of neural networks with mixed delays. For neural networks with <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${n}$ </tex-math></inline-formula> neurons, <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${m}$ </tex-math></inline-formula> -dimensional input vectors, and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${2k}$ </tex-math></inline-formula> -valued activation functions, the autoassociative memories have <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${(2k)^{n}}$ </tex-math></inline-formula> storage capacities and heteroassociative memories have min <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${\{(2k)^{n},(2k)^{m}\}}$ </tex-math></inline-formula> storage capacities. That is, the storage capacities of designed associative memories in this article are obviously higher than the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${2^{n}}$ </tex-math></inline-formula> and min <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${\{2^{n},2^{m}\}}$ </tex-math></inline-formula> storage capacities of the conventional ones. Three examples are given to support the theoretical results.

Topics & Concepts

Associative propertyBidirectional associative memoryContent-addressable memoryComputer scienceArtificial neural networkSpurious relationshipUniquenessAsynchronous communicationRecurrent neural networkEquilibrium pointStability (learning theory)Content-addressable storageOverhead (engineering)Exponential stabilityMathematicsArtificial intelligencePure mathematicsDifferential equationNonlinear systemMachine learningComputer networkOperating systemMathematical analysisPhysicsQuantum mechanicsNeural Networks Stability and SynchronizationAdvanced Memory and Neural ComputingNeural Networks and Applications
Analysis and Design of Multivalued High-Capacity Associative Memories Based on Delayed Recurrent Neural Networks | Litcius