Mutual Information Based Initialization of Forward-Backward Search for Feature Selection in Regression Problems
Pure feature selection, where variables are chosen or not to be in the training data set, still remains as an unsolved problem, especially when the dimensionality is high. Recently, the Forward-Backward Search algorithm using the Delta Test to evaluate a possible solution was presented, showing a good performance. However, due to the locality of the search procedure, the initial starting point of the search becomes crucial in order to obtain good results. This paper presents new heuristics to find a more adequate starting point that could lead to a better solution. The heuristic is based on the sorting of the variables using the Mutual Information criterion, and then performing parallel local searches. These local searches provide an initial starting point for the actual parallel Forward-Backward algorithm.
KeywordsFeature Selection Local Search Mutual Information Time Series Prediction Sorting Scheme
Unable to display preview. Download preview PDF.
- 1.Eirola, E., Liitiäinen, E., Lendasse, A., Corona, F., Verleysen, M.: Using the Delta Test for Variable Selection. In: ESANN 2008, European Symposium on Artificial Neural Networks, Bruges, Belgium (April 2008)Google Scholar
- 7.Guillen, A., Rojas, I., Rubio, G., Pomares, H., Herrera, L.J., Gonzalez, J.: A new interface for MPI in matlab and its application over a genetic algorithm. In: Lendasse, A. (ed.) Proceedings of the European Symposium on Time Series Prediction, pp. 37–46 (2008), http://atc.ugr.es/~aguillen