Task-driven deep transfer learning for image classification

Transfer learning tends to be a powerful tool that can mitigate the divergence across different domains through knowledge transfer. Recent research efforts on transfer learning have exploited deep neural network (NN) structures for discriminative feature representation to better tackle cross-domain disparity. However, few of these techniques are able to jointly learn deep features and train a classifier in a unified transfer learning framework. To this end, we design a task-driven deep transfer learning framework for image classification, where the deep feature and classifier are obtained simultaneously for optimal classification performance. Therefore, the proposed deep structure can generate more discriminative features by using the classifier performance as a guide. Furthermore, the classifier performance is increased since it is optimized on a more discriminative deep feature. The developed supervised formulation is a task-driven scheme, which will provide better learned features for the classification task. By giving pseudo labels for target data, we can facilitate the knowledge transfer from source to target through the deep structures. Experimental results witness the superiority of our proposed algorithm by comparing with other ones.

1. Introduction

In typical pattern recognition problems, there is always a situation that we have plenty of unlabeled data while there are limited or even no labeled data for training from the test (target) domain. Transfer learning [1] has been demonstrated as a promising technique to address such difficulty by borrowing knowledge from other well-learned source domains, which might lie in a different distribution than the target domain. Recent research on transfer learning have witnessed appealing performance by seeking a common feature space where knowledge from source can be well transferred to assist the recognition task in target domain [2], [3], [4], [5], [6], [7]. Fig. 1.

Illustration of our proposed -layer coupled deep neural network (here ). Deep structures are built to learn deep features for source and target domains , which share the same networks weights . A pair-wise constraint is developed to couple the similar pair of source and target in the layer to transfer knowledge. Moreover, a classifier is jointly trained on source data, where is the label matrix of source data.