Learning Multiple Categories on Deep Convolution Networks

TL;DR

This paper demonstrates that deep convolutional networks can effectively decompose large recognition tasks into smaller subtasks, learning them simultaneously, which explains their success on big datasets like ImageNet.

Contribution

It introduces a method for training deep convolutional networks on multiple categories simultaneously, showing near-equivalent performance to separate networks for each subtask.

Findings

Networks can decompose complex tasks into smaller ones effectively.

Using task-specific labels improves recognition performance.

Performance on combined tasks approaches that of separate networks.

Abstract

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.