A novel approach for integrating feature and instance selection
Jerffeson Souza, Rafael Augusto Ferreira do Carmo, Gustavo Augusto Lima de Campos · 2008
As important machine learning problems, feature and instance selection have faced relevant improvements in the quality of the algorithms that solve them individually. However, little work has been done to implement ways to solve them simultaneously. In this paper, we introduce an algorithm that combines solutions for both problems, using a simple adaptation of the simulated annealing metaheuristic. Our empirical evaluation shows that, when time constraints are present, our algorithm outperforms other similar strategies.