Impact of a DCT-driven Loss in Attention-based Knowledge-Distillation for Scene Recognition. (arXiv:2205.01997v1 [cs.CV])

Web: http://arxiv.org/abs/2205.01997

Knowledge Distillation (KD) is a strategy for the definition of a set of
transferability gangways to improve the efficiency of Convolutional Neural
Networks. Feature-based Knowledge Distillation is a subfield of KD that relies
on intermediate network representations, either unaltered or depth-reduced via
maximum activation maps, as the source knowledge. In this paper, we propose and
analyse the use of a 2D frequency transform of the activation maps before
transferring them. We pose that\textemdash by using global image cues rather
than …

arxiv attention cv distillation impact knowledge knowledge-distillation loss