Improved Classifier Chain Method Based on Particle Swarm Optimization and Genetic Algorithm for Multilabel Classification Problem

Classifier chain (CC) is one of the most important MLC methods often applied in multi- labeling tasks. However, the standard CC model fails to recognize distinct labels sequence order in the chain. In this work, a new method is proposed to optimize the chaining order in CC for improved classification performance. The proposed method is based on particle swarm optimization and genetic algorithm. Genetic operators are integrated with the standard PSO algorithm for finding the global best solution representing an optimized label sequence order in the chain. In the experiment, 6 datasets and 8 evaluation metrics are employed. The new method achieved the overall best results including 98.66% accuracy and 97.49% exact match among other metrics applied.