A Multiple Pheromone Algorithm for Cluster Analysis

Ant colony optimisation algorithms (ACO) work via a process called stigmergy in which ants deposit pheromone traces in order to influence foraging patterns. Pheromone traces are picked up and followed by other ants but they evaporate over time. Paths with more pheromone will survive longer and have a higher chance of getting followed and reinforced whilst weaker traces simply fade away. The premise behind the proposed Multiple Pheromone Algorithm for Cluster Analysis (MPACA) is that ants detect individual features of objects in space and deposit pheromone traces that guide towards these features. Each ant starts off by looking for a particular feature but they can combine with ants looking for other features if the match of their paths is above a given threshold. This enables ants to detect and deposit pheromone corresponding to feature combinations and provides the colony with more powerful cluster analysis and classification tools. The basic elements of MPACA are that: (i) at the start of the learning process, every object has at least one ant assigned to it for each feature; (ii) each ant searches for other objects with a matching feature value; (iii) a pheromone is laid down whenever an ant has found an object with a matching feature; (iv) if ants detecting different features find their paths are matching above a certain level, they will combine and start looking for the conjunction of features; and (v) ants become members of the same colony when the population density of ants in the area is above a threshold value. This paper explains the algorithm and explores its potential effectiveness for cluster analysis.