Utilizing Architecture Aspects for in Data Mining for Computer System Design

Abstract

Data mining has been a popular technique. It has been for many applications in many areas nowadays. In this chapter, we are interested in utilizing architecture features and apply data mining techniques for a computer design. Since data mining requires lot of collected data for building models, the relevant data needs to be properly generated. We demonstrate the applications and their methodology starting from data set generation, feature extraction, modeling and evaluations. Important characteristics of the architecture are considered for data set generation and feature extractions: particularly, the instruction set, and memory access pattern features. The chapter utilizes these features given with observations for building the models for cache prediction, branch prediction, and malware detection.