Finding groups in data: Cluster analysis with ants

Enhancing the effectiveness of Ant Colony Decision Tree algorithms by co-learning

“…One of the first works on this subject focuses on data clustering [24], and the suggested method has been applied to data mining in the area of decision tree construction with good results [2,3].…”

“…Because of bi-criteria evaluation of decision trees, we propose to analyze the Pareto front constructed by solutions. An analysis of the decision tree evaluation concerning the two criteria in the ACDT algorithm is specially dedicated to parameter and the way the decision tree growth is calculated: (7), (23), (24) and (25). The value of parameter is established as 1.0, while the classification accuracy is determined by formula (3).…”

A novel ant-based clustering algorithm using the kernel method

“…Fast-moving ants form rough clusters on a large scale, while slow ants group objects on a smaller scale by dropping objects with more accuracy. Using ants with short-term memory [2,16,17,36]. An ant stores two objects in its short-term memory: visited places and visited objects.…”

“…Making some important parameters adaptive, such as the neighbourhood radius r and the scale factor a [2,14,15,36,39].…”

“…It takes long time to go from the inception of the algorithm to the moment when the rough clusters are created. Commonly, 100,000 iterations are needed for ant-based clustering algorithms [2]. Improving the algorithm's clustering quality In essence, ant-based clustering algorithms are distance-based because the similarity of the objects is computed by Euclidean distance or Cosine distance.…”