Tolerance-Based Approach to Audio Signal Classification

In this paper, we propose a supervised learning algorithm to classify audio signals based on the tolerance near sets model (TNS). In the TNS method, tolerance classes are directly induced from the data set using the tolerance level epsilon and a distance function. Preliminary experiments with an audio signal data set show promising results in terms of the accuracy of classifier. Overall, TCL is able to demonstrate similar performance in terms of accuracy with Fuzzy IDT algorithm [1] and comparable performance with a rough set based classifier as well as classical machine learning algorithms based on decision trees, rules, bayesian learning and support vector machines.