Sparse extreme learning machine using privileged information for classification

Abstract

In human learning process, teachers play an important role of transferring knowledge and providing some additional information for better understanding. In machine learning, the role of the teacher has been mostly ignored. Introduction of a new learning paradigm named Learning Using Privileged Information (LUPI) includes the elements of human teaching in machine learning. This paper proposes a learning method using privileged information based on a Sparse Extreme Learning Machine (ELM). Our method aims to improve the classification performance by reducing the testing error. Experimental results show improvements in training accuracy and testing error reduction over the classical methods in classification.