A topological and temporal correlator network for spatiotemporal pattern learning, recognition, and recall

In this paper, we describe the design of an artificial neural network for s patiotemporal pattern recognition and recall. This network has a five-layer ed architecture and operates in two modes: pattern learning and recognition mode, and pattern recall mode, In pattern learning and recognition mode, t he network extracts a set of topologically and temporally correlated featur es from each spatiotemporal input pattern based on a variation of Kohonen's self-organizing maps. These features are then used to classify the input i nto categories based on the fuzzy ART network. In the pattern recall mode, the network can reconstruct any of the learned categories when the appropri ate category node is excited or probed, The network performance was evaluat ed via computer simulations of time-varying, two-dimensional and three-dime nsional data. The results show that the network is capable of both recognit ion and recall of spatiotemporal data in an on-line and self-organized fash ion. The network can also classify repeated events in the spatiotemporal in put and is robust to noise in the input such as distortions in the spatial and temporal content.