Langevin Cooling for Domain Translation
This work addresses a specific issue in domain translation for applications like image and language processing, offering an incremental improvement over existing methods.
The paper tackled the problem of domain translation where existing methods often fail on fringe samples in low-density areas, and proposed Langevin Cooling (L-Cool) to improve performance by moving such samples towards high-density regions, showing enhancements in image and language translation tasks.
Domain translation is the task of finding correspondence between two domains. Several Deep Neural Network (DNN) models, e.g., CycleGAN and cross-lingual language models, have shown remarkable successes on this task under the unsupervised setting---the mappings between the domains are learned from two independent sets of training data in both domains (without paired samples). However, those methods typically do not perform well on a significant proportion of test samples. In this paper, we hypothesize that many of such unsuccessful samples lie at the fringe---relatively low-density areas---of data distribution, where the DNN was not trained very well, and propose to perform Langevin dynamics to bring such fringe samples towards high density areas. We demonstrate qualitatively and quantitatively that our strategy, called Langevin Cooling (L-Cool), enhances state-of-the-art methods in image translation and language translation tasks.