Knowledge as Priors: Cross-Modal Knowledge Generalization for Datasets without Superior Knowledge

Abstract

Cross-modal knowledge distillation deals with transferring knowledge from a model trained with superior modalities (Teacher) to another model trained with weak modalities (Student). Existing approaches require paired training examples exist in both modalities. However, accessing the data from superior modalities may not always be feasible. For example, in the case of 3D hand pose estimation, depth maps, point clouds, or stereo images usually capture better hand structures than RGB images, but most of them are expensive to be collected. In this paper, we propose a novel scheme to train the Student in a Target dataset where the Teacher is unavailable. Our key idea is to generalize the distilled cross-modal knowledge learned from a Source dataset, which contains paired examples from both modalities, to the Target dataset by modeling knowledge as priors on parameters of the Student. We name our method "Cross-Modal Knowledge Generalization" and demonstrate that our scheme results in competitive performance for 3D hand pose estimation on standard benchmark datasets.

Two-Branch Attention Learning for Fine-Grained Class Incremental Learning

“…In LwF [13], knowledge distillation regularization term is first introduced into the loss function to retain the knowledge learnt from the previous training data. Knowledge distillation refers to distilling knowledge from a model of a cumbersome teacher and infusing it to a model of a light student which applied extensively in teaching [23][24][25][26] and can contribute to a generalization of model [27][28][29]. As shown in [9], LwF prefers to process new classes in the inference phase.…”