Improving pattern recognition of electronic nose data with time-delay neural networks

A review of unsupervised feature learning and deep learning for time-series modeling

“…It has been used for identifying the smell of spices (Zhang et al, 2003), ternary mixtures (Vito et al, 2007), optimum fermentation time for black tea (Bhattacharya et al, 2008), and vintages of wine (Yamazaki et al, 2001). An RNN have been used for odour localization with a mobile robot (Duckett et al, 2001).…”

Neural networks-integrated metal oxide-based artificial olfactory system for meat spoilage identification

“…They concluded that neural networks performed better than multivariate statistical-based method (cluster analysis). Zhang et al (2003) also demonstrated the superior performance of ANN-based data analysis techniques over statistical-based data analysis techniques reporting a 41% increase in classification accuracy while using ANN-based time-delay neural network over statistical-based discriminant function analysis to discriminate between four spices.…”

“…Artificial neural networks (ANNs) are dynamic and self-adapting systems, which closely resemble the process of human learning and have been widely utilized to handle nonlinear data processing from electronic nose systems (Llobet et al, 2004;Parpinello et al, 2007). Back propagation neural network (BPNN), support vector machines (SVM), adaptive logic network (ALN), radial basis function network (RBFN), fuzzy ARTMAP, self organizing map (SOM) network and time-delay neural network are some of the ANN architectures that have been applied to process electronic nose data (Qu et al, 2001;Dutta et al, 2002Dutta et al, , 2006Zhang et al, 2003;Brudzewski et al, 2004;Llobet et al, 2004;Yu et al, 2008). Selection of the appropriate neural network architecture involves a variety of factors including sample size, number of inputs, number of out puts, hidden layers, training method (supervised or unsupervised) and purpose of the network (classification or prediction) with the overall goal to find an appropriate input-output relation that minimizes the prediction error (Yang et al, 1998;Agatonovic-Kustrin and Beresford, 2000).…”

Determination of Quality Attributes of Blue Crab (Callinectes sapidus) Meat by Electronic Nose and Draeger-Tube Analysis

“…Therefore, a neural network can "learn" as data is imputed in, whereas a pattern recognition technique simply looks for patterns. The use of neural networks significantly improved pattern recognition for electronic nose systems (Zhang et al, 2003) and was found to be effective for quantifying different spice mixture compositions (Zhang et al, 2005). Due to time constraints, this technique was not applied to this data set.…”