Learning Multiple Categories on Deep Convolution Networks
This provides insights into the scalability of deep learning for multi-category recognition, but it is incremental as it builds on existing network architectures.
The paper investigates why deep convolutional networks perform well on large-scale recognition tasks by showing they can decompose complex tasks into smaller subtasks and learn them simultaneously, achieving performance close to using separate networks for each subtask.
Deep convolution networks have proved very successful with big datasets such as the 1000-classes ImageNet. Results show that the error rate increases slowly as the size of the dataset increases. Experiments presented here may explain why these networks are very effective in solving big recognition problems. If the big task is made up of multiple smaller tasks, then the results show the ability of deep convolution networks to decompose the complex task into a number of smaller tasks and to learn them simultaneously. The results show that the performance of solving the big task on a single network is very close to the average performance of solving each of the smaller tasks on a separate network. Experiments also show the advantage of using task specific or category labels in combination with class labels.