Energy-Efficient Sound Environment Classifier for Hearing Aids Based on Multi-objective Simulated Annealing Programming

A methodology for designing classifiers through multicriteria metaheuristics is introduced. The purpose of the new method is to jointly optimize the expected total classification cost and the energy consumption of the device the classifier is implemented on. A numerical study is provided, where different alternatives are implemented on a hearing aid. This aid is capable of automatically classifying the acoustic environment that surrounds the user and choosing the parameters of the amplification that are best adapted to the user's comfort. The proposed method attains relevant improvements in energy consumption with small to negligible loss in classification accuracy with respect to a selection of algorithms.