Hierarchical multi-label classification using local neural networks

• We incrementally train a multi-layer perceptron for each level of the classification hierarchy.
• Predictions made in a level are used as inputs to the neural network associated to the next level.
• Local information learned in a level was useful to learn a neural network in the next level.
• We discussed about the smaller number of predictions made by our method in the deeper hierarchical levels.
• Experimental analysis show that our method obtains better or competitive results.

Hierarchical multi-label classification is a complex classification task where the classes involved in the problem are hierarchically structured and each example may simultaneously belong to more than one class in each hierarchical level. In this paper, we extend our previous works, where we investigated a new local-based classification method that incrementally trains a multi-layer perceptron for each level of the classification hierarchy. Predictions made by a neural network in a given level are used as inputs to the neural network responsible for the prediction in the next level. We compare the proposed method with one state-of-the-art decision-tree induction method and two decision-tree induction methods, using several hierarchical multi-label classification datasets. We perform a thorough experimental analysis, showing that our method obtains competitive results to a robust global method regarding both precision and recall evaluation measures.