A Bidirectional Hetero-Associative Memory for True-Color Patterns

Abstract

Classical bidirectional associative memories (BAM) have poor memory storage capacity, are sensitive to noise, are subject to spurious steady states during recall, and can only recall bipolar patterns. In this paper, we introduce a new bidirectional hetero-associative memory model for true-color patterns that uses the associative model with dynamical synapses recently introduced in Vazquez and Sossa (Neural Process Lett, Submitted, 2008). Synapses of the associative memory could be adjusted even after the training phase as a response to an input stimulus. Propositions that guarantee perfect and robust recall of the fundamental set of associations are provided. In addition, we describe the behavior of the proposed associative model under noisy versions of the patterns. At last, we present some experiments aimed to show the accuracy of the proposed model with a benchmark of truecolor patterns.

Fault Tolerance Technique Using Bidirectional Hetero-Associative Memory for Self-Reconfigurable Programmable Matter

“…Bidirectional Hetero-Associative Memory (BHAM) [15], [8] is a supervised neural network model with the ability to memorize the associations between input and output patterns without need of continuous learning. These associations will be stored only once and recalled by the BHAM for patterns recognition.…”

“…2) Number of packets pairs association: Our objective in this experimentation is to show the behavior of the BHAM network when the number of associated packets pairs increases while the size of the network (m, n) remains unchanged. In Figure 10 we varied this number between 3 and 5 for the same size (15,30). It is shown in this result that when the number of associated packets pairs increases, the percentage of corrected packets decreases but not drastically.…”

Existence and Exponential Stability of Multiple Periodic Solutions for a Multidirectional Associative Memory Neural Network

“…Bidirectional associative memory (BAM) neural networks, which were first introduced by Kosto [15], can realize hetero-association. These neural networks have attracted a lot of attention of many researchers [2,4,3,5,6,11,12,16,17,19,18,23,24,26,27,[29][30][31] and found many important applications in pattern recognition, image processing, associative memory and optimization problems [14,22,20,25]. However, for the human brain associative memory, besides these basic functions, there are some other more complicated associative memory functions, such as many-to-many association, which associates a number of related events with one hit or several features.…”