Interpretable Piecewise Linear Classifier

In this study we propose a new ensemble model composed of several linear perceptrons. The objective of this study is to build a piecewise-linear classifier that is not only competitive to Multilayer Perceptrons(MLP) in generalization performance but also interpretable in the form of human-comprehensible rules. We present a simple competitive training method that allows the ensemble to effectively divide a given training space into several sub-spaces on the basis of so called ”confidence value”, and train each module to obtain a linear rule within the allocated sub-space. The linearity of the ensemble’s module significantly simplifies the rule extraction process.

• Type: Conference
• Year: 2007
• Where: ICONIP
• Authors: Pitoyo Hartono
• Keywords: ICONIP 2007, Information Technology, Linear Perceptrons, Rule Extraction Process, Simple Competitive Training