Nonnegative Tensor Decomposition Via Collaborative Neurodynamic Optimization

TL;DR

This paper proposes a collaborative neurodynamic neural network approach combined with particle swarm optimization to compute nonnegative tensor decompositions, with proven convergence and demonstrated effectiveness on various datasets.

Contribution

It introduces a novel neurodynamic model for nonnegative tensor decomposition that integrates PSO for improved optimization and provides convergence analysis.

Findings

Effective in computing nonnegative CPD on real-world datasets

Convergence and stability of the proposed models are validated

Neurodynamic approach outperforms traditional methods

Abstract

This paper introduces a novel collaborative neurodynamic model for computing nonnegative Canonical Polyadic Decomposition (CPD). The model relies on a system of recurrent neural networks to solve the underlying nonconvex optimization problem associated with nonnegative CPD. Additionally, a discrete-time version of the continuous neural network is developed. To enhance the chances of reaching a potential global minimum, the recurrent neural networks are allowed to communicate and exchange information through particle swarm optimization (PSO). Convergence and stability analyses of both the continuous and discrete neurodynamic models are thoroughly examined. Experimental evaluations are conducted on random and real-world datasets to demonstrate the effectiveness of the proposed approach.