2017 IEEE International Conference on Wireless Communications, Signal Processing and Networking (WiSPNET 2017)

208 / 2017-01-09 19:09:20

Constructing a Linear discrete system in Kernel space as a supervised classifier

1857,12347,12348,5604,12349

The pattern recognition techniques involve feature extraction from the data, dimensionality reduction (like PCA, LDA, K-LDA, etc) and constructing a classifier (NN, NM, SVM, etc.) using the training set and validating the constructed classifier using the testing set. The usage of digital signal processing (DSP) techniques in pattern recognition is always limited to the feature extraction stage such as collecting the Fourier, wavelet co-efficients, HMM, GMM, etc. In this paper we explore the usage of classical DSP techniques like convolution, FIR filter to construct the classifier and are compared with the state of the art techniques. The proposed technique paves the alternative way to construct a classifier that is helpful for Big data analysis.