Blockout: Dynamic Model Selection for Hierarchical Deep Networks
This addresses the need for more efficient and adaptive hierarchical deep networks in image classification, offering a novel regularization approach that is incremental over existing methods like Dropout.
The paper tackles the problem of learning shared image representations in deep networks by proposing Blockout, a method for dynamic model selection that simultaneously learns hierarchical architectures and parameters, resulting in improved classification accuracy, better regularization, faster training, and emergent hierarchical structures on CIFAR and ImageNet datasets.
Most deep architectures for image classification--even those that are trained to classify a large number of diverse categories--learn shared image representations with a single model. Intuitively, however, categories that are more similar should share more information than those that are very different. While hierarchical deep networks address this problem by learning separate features for subsets of related categories, current implementations require simplified models using fixed architectures specified via heuristic clustering methods. Instead, we propose Blockout, a method for regularization and model selection that simultaneously learns both the model architecture and parameters. A generalization of Dropout, our approach gives a novel parametrization of hierarchical architectures that allows for structure learning via back-propagation. To demonstrate its utility, we evaluate Blockout on the CIFAR and ImageNet datasets, demonstrating improved classification accuracy, better regularization performance, faster training, and the clear emergence of hierarchical network structures.