Multi-label Classification via Adaptive Resonance Theory-based Clustering
Naoki Masuyama, Yusuke Nojima, Chu Kiong Loo, Hisao Ishibuchi
TL;DR
The paper tackles continual learning for multi-label classification where the label set can grow over time and data arrive sequentially. It proposes MLCA, an ART-based CIM clustering approach that integrates Bayesian label probability estimates, plus two variants MLCA-I and MLCA-C to handle heterogeneous attributes. Empirical results on synthetic and 16 real-world datasets show MLCA achieves competitive accuracy while effectively accumulating knowledge through new nodes, enabling continual and robust learning in non-stationary environments. The work offers scalable, prototype-based classification without heavy pre-processing, with clear avenues for handling concept drift and mixed data types in future work.
Abstract
This paper proposes a multi-label classification algorithm capable of continual learning by applying an Adaptive Resonance Theory (ART)-based clustering algorithm and the Bayesian approach for label probability computation. The ART-based clustering algorithm adaptively and continually generates prototype nodes corresponding to given data, and the generated nodes are used as classifiers. The label probability computation independently counts the number of label appearances for each class and calculates the Bayesian probabilities. Thus, the label probability computation can cope with an increase in the number of labels. Experimental results with synthetic and real-world multi-label datasets show that the proposed algorithm has competitive classification performance to other well-known algorithms while realizing continual learning.