A Scale and Translation Invariant Approach for Early Classification of Spatio-Temporal Patterns Using Spiking Neural Networks

摘要

This paper addresses the problem of encoding and classifying spatio-temporal patterns, which are typical for human actions or gestures. The proposed method has the following main contributions: (i) it requires a very small number of training examples, (ii) it accepts variable sized input patterns, (iii) it is invariant to scale and translation, and (iv) it enables early recognition, from only partial information of the pattern. The underlying representation employed is a spiking neural network with axonal conductance delay. We designed a novel approach for mapping spatio-temporal patterns to spike trains, which are used to stimulate the network. The pattern features emerge in the network as a result of this stimulation in the form of polychronous neuronal groups, which are used for classification. The proposed method is validated on a set of gestures representing the digits from \(0\) to \(9\), extracted from video data of a human drawing the corresponding digits. The paper presents a comparison with several other standard pattern recognition approaches. The results show that the proposed approach significantly outperforms these methods, it is invariant to scale and translation, and it has the ability to recognize patterns from only partial information.