A Hierarchical Bayes-Based Evolutionary Ensemble Classification Algorithm

Regarding a multi-label classification task, the different similarities between targeted categories are commonly overlooked in front of deep neural network models. Never-theless, decomposing the large-scale multi-label classification problem into a series of hierarchical sub-problems based on their similarity information can reduce the problem dimension and result in lower computational cost with competitive performance. This paper proposed a Hierarchical Bayes-based Evolutionary Ensemble (HBEE) classification algorithm that computes and utilises our new data-driven posterior-based class similarity to evolve a tree of weak classifiers. The posterior information is gathered from a reduced Bayes theorem, which is insensitive to imbalanced data amount and imbalanced inter-class similarities. Instead of doing gradient descent optimization on large scale parameters, a gradient-free optimization method, genetic technique, is adopted for a series of weak classifier's ensemble decision, which is extremely useful in industry when large scale gradient optimization method is not feasible.