Data Oriented Approximate K-Nearest Neighbor Classifier for Touch Modality Recognition

Approximate computing techniques offer a promising solution to reduce the hardware complexity and power consumption imposed when embedding machine learning algorithms. The reduction comes at the cost of some performance degradation. This paper presents an approximate machine learning classifier for touch modality recognition. Experimental results demonstrate that the use of software level approximation techniques reduce the execution time and memory usage up to 38% and 55% respectively, at the cost of accuracy loss less than 10% for the target application.