Robust Sparse Learning Based Sensor Array Optimization for Multi-feature Fusion Classification

In this paper, we propose a robust sensor array optimization method based on sparse learning for multi-feature fusion data classification. The proposed approach contains three key characteristics. First, it considers the intrinsic group structure among features by combining an l(F,1) norm regularizer design and least squares regression framework. Second, in sensor selection, insignificant feature groups can be eliminated by grouped row sparse coefficients generated by the model, while the epsilon-dragging trick is introduced to improve the classification ability. Third, an efficient alternating iteration algorithm is presented to optimize the convex objective function. The results compared with the other classical methods on gas sensor array data sets demonstrate that the proposed method can effectively reduce the number of sensors with higher classification accuracy.

Automated summary

This paper proposes a robust sensor array optimization method based on sparse learning for multi-feature fusion data classification. The method considers the intrinsic group structure among features and eliminates insignificant feature groups through sparse coefficients generated by the model. An efficient alternating iteration algorithm is presented to optimize the objective function. Experimental results demonstrate that the proposed method effectively reduces the number of sensors with improved classification accuracy.