Deeply Coupled Graph Structured Autoencoder for Domain Adaptation
Angshul Majumdar · 2019
This work introduced graph regularization to the framework of deep coupled autoencoders. The ensuing domain adaptation tool has been used on a variety of application areas -- image super-resolution, translingual document retrieval and multimodal information retrieval. In each case, the proposed method outperforms other coupled representation learning formulations and application specific approaches.