The Multiverse Loss for Robust Transfer Learning
This work tackles the problem of improving transfer learning robustness for deep learning applications, though it appears incremental as it builds on existing transfer learning methods.
The paper addresses the underutilization of dimensionality in transferred representations by proposing a method to learn multiple orthogonal classifiers in the source domain, which leads to a reduced rank representation that supports more discriminative directions. Experimental results on CIFAR-100 and LFW demonstrate its effectiveness.
Deep learning techniques are renowned for supporting effective transfer learning. However, as we demonstrate, the transferred representations support only a few modes of separation and much of its dimensionality is unutilized. In this work, we suggest to learn, in the source domain, multiple orthogonal classifiers. We prove that this leads to a reduced rank representation, which, however, supports more discriminative directions. Interestingly, the softmax probabilities produced by the multiple classifiers are likely to be identical. Experimental results, on CIFAR-100 and LFW, further demonstrate the effectiveness of our method.