Self-adaptive ant colony algorithm for attributes reduction

The attributes reduction (AR) and their values are one of the highlight of rough set theory. Decision table can be simplified effectively by critical attributes and their values. But this problem is an NP-hard problem. On terms of the similarity and difference between TSP (travel salesman problem) and AR, the ant colony algorithm (ACA) is applied to solve AR problem, and an self-adaptive ant colony algorithm (SAACA) is proposed, which is improved from ACA through modifying the pheromone updating rule and the transition rule by introducing evenness of solution and interests into it in order to reduce computing time and avoid to stagnation behavior of basic ACA. Simulation results show that the AACA can settle the contradictory between convergence speed and stagnation behavior efficiently and is very suitable for solving AR.