Experimental evaluation of latency coding for gas recognition

Abstract

Commercial gas recognition systems use advanced computationally intensive signal processing/pattern recognition algorithms to identify gases and discriminate between them. This severely impacts on the size and cost of such systems but also limits their large-scale deployment. Biologically-inspired gas recognition schemes have the potential to greatly simplify the task of gas recognition, enabling the advent of low cost and low power miniature gas systems. In this paper, we present an experimental evaluation of bio-inspired latency coding for gas recognition. The performance of this bio-inspired approach was evaluated against four commonly used pattern recognition algorithms, namely K Nearest Neighbors (KNN), neural networks (Multi-Layer Perceptron (MLP), Radial Basis Function (RBF)) and density models (Gaussian Mixture Models (GMM). Reported experimental results suggest that latency coding could perform as well if not better than more computationally intensive pattern recognition techniques.