RST: A Connectionist Architecture to Deal with Spatiotemporal Relationships

Abstract

In the past decade, connectionism has proved its efficiency in the field of static pattern recognition. The next challenge is to deal with spatiotemporal problems. This article presents a new connectionist architecture, RST (réseau spatio temporel [spatio temporal network]), with such spatiotemporal capacities. It aims at taking into account at the architecture level both spatial relationships (e.g., as between neighboring pixels in an image) and temporal relationships (e.g., as between consecutive images in a video sequence). Concerning the spatial aspect, the network is embedded in actual space (two) or three-dimensional-, the metrics of which directly influence its structure through a connection distribution function. For the temporal aspect, we looked toward biology and used a leaky-integrator neuron model with a refractory period and postsynaptic potentials. The propagation of activity by spatiotemporal synchronized waves enables RST to perform motion detection and localization in sequences of video images.

A neuro-fuzzy monitoring system

“…The detection tool is based on the recurrent RBF neural network (Zemouri et al, 2001(Zemouri et al, , 2003. The RRBF neural network considers the time as an internal representation (Chappelier and Grumbach, 1998;Elman, 1990;Zemouri et al, 2003). The dynamic aspect is obtained using an additional self-connection of input neurons with a sigmoid activation function.…”

Sequence Learning

“…For instance, RST (Chappelier & Grumbach, 1998) which has been applied to finding the moving part of an image, uses properties 2 and 3 above 3 .…”

“…The approach of (Chappelier & Grumbach, 1998) consists of embedding spatial dimensions into the network and integrating time at both neuron and 16 technically: "to win". The approach of (Chappelier & Grumbach, 1998) consists of embedding spatial dimensions into the network and integrating time at both neuron and 16 technically: "to win".…”

Time in Connectionist Models

“…For instance, RST (Chappelier and Grumbach 1998) which has been applied to finding the moving part of an image, uses properties 2 and 3 above 3 .…”

“…RST. The approach of Chappelier and Grumbach (1998) consists of embedding spatial dimensions into the network and integrating time at both neuron and connection levels. The aim is to take both spatial relationships (e.g.…”