IEEE Transactions on Pattern Analysis and Machine Intelligence
Multiobjective evolutionary algorithms: a comparative case studyand the strength Pareto approach
IEEE Transactions on Evolutionary Computation
| Hi-index | 0.00 |
High-dimensional data often threatens the performance of classification algorithms. We propose a two-step approach for dealing with high-dimensional data. In the first step, features are arranged into bins, where each bin corresponds to a much smaller sub-space of features. In the second step, classifiers are independently applied to the set of features within each sub-space, and their results are then aggregated. We consider slicing a space Rd into smaller subspaces as a multi-objective search problem, which can be solved by evolutionary algorithms. We performed a systematic evaluation using three classification algorithms on high-dimensional data.