Scalable Multitask Representation Learning for Scene Classification

Maksim Lapin, Bernt Schiele, Matthias Hein; Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2014, pp. 1434-1441

The underlying idea of multitask learning is that learning tasks jointly is better than learning each task individually. In particular, if only a few training examples are available for each task, sharing a jointly trained representation improves classification performance. In this paper, we propose a novel multitask learning method that learns a low-dimensional representation jointly with the corresponding classifiers, which are then able to profit from the latent inter-class correlations. Our method scales with respect to the original feature dimension and can be used with high-dimensional image descriptors such as the Fisher Vector. Furthermore, it consistently outperforms the current state of the art on the SUN397 scene classification benchmark with varying amounts of training data.