Hetero Chaotic Associative Memory for Successive Learning with Multi-Winners Competition

In this paper, we propose a Hetero Chaotic Associative Memory for Successive Learning with Multi-Winners Competition (HCAMSL-MW). The proposed model is based on a Hetero Chaotic Associative Memory for Successive Learning (HCAMSL) and the Multi Winners Self-Organizing Neural Network (MWSONN). In most of the conventional neural network models, the learning process and the recall process are divided, and therefore they need all information to learn in advance. However, in the real world, it is very difficult to get all information to learn in advance. So we need the model whose learning and recall processes are not divided. As such model, although some models have been proposed, most of them can deal with only auto-associations. In contract, although the conventional HCAMSL can deal with hetero-associations, its storage capacity is small. In the proposed HCAMSL-MW, the storage capacity is improved by the internal pattern generation based on the multi-winners competition. We carried out a series of computer experiments and confirmed that the storage capacity of the proposed HCAMSLMW is larger than that of the conventional HCAMSL.