A Spatio-Temporal Associative Memory Architecture Using Chaotic Masking Technique for Complex Sequence Storage and Retrieval

A spatio-temporal associative memory neural network architecture based on hetero-associative memory neural network and chaotic masking techniques is proposed. By using chaotic masking techniques for preprocessing and multiple hetero-associative memory neural networks, the process of complex sequences avoids many problems like confusion and gets easier. The experiments show that this architecture is capable of storing and retrieving complex spatio-temporal patterns, where some spatial patterns are highly-correlated and even repetitive while other related works can not do.