An Adaptive Ant System using Momentum Least Mean Square Algorithm
Abhishek Paul, Sumitra Mukhopadhyay · 2013
paper, a novel model has been proposed for pheromone updation of the Ant-System, entitled as Momentum Adaptive Ant System (MAAS). MAAS exploits the properties of Adaptive Filters. The proposed algorithm is implemented using momentum-LMS (Least Mean Square) based algorithm. It imparts information about the previous occurrence of the system so as to keep the system active even in the region close to the minimum (i.e., minimum optimal) solution. MAAS modifies its properties in accordance to the requirement of surrounding realm and for the betterment of its performance in dynamic environment. The proposed algorithm overcomes stagnation and offers better searching capability. Also it helps the ants (i.e., co-operating agents) not to get stuck at local optima. The results of experimental study are well described and it establishes the usefulness of the new strategy. Proposed algorithm shows effective performance when applied to the Traveling Salesman Problem (TSP).