arXiv Analytics

Sign in

arXiv:1811.03233 [cs.LG]AbstractReferencesReviewsResources

Knowledge Transfer via Distillation of Activation Boundaries Formed by Hidden Neurons

Byeongho Heo, Minsik Lee, Sangdoo Yun, Jin Young Choi

Published 2018-11-08Version 1

An activation boundary for a neuron refers to a separating hyperplane that determines whether the neuron is activated or deactivated. It has been long considered in neural networks that the activations of neurons, rather than their exact output values, play the most important role in forming classification friendly partitions of the hidden feature space. However, as far as we know, this aspect of neural networks has not been considered in the literature of knowledge transfer. In this paper, we propose a knowledge transfer method via distillation of activation boundaries formed by hidden neurons. For the distillation, we propose an activation transfer loss that has the minimum value when the boundaries generated by the student coincide with those by the teacher. Since the activation transfer loss is not differentiable, we design a piecewise differentiable loss approximating the activation transfer loss. By the proposed method, the student learns a separating boundary between activation region and deactivation region formed by each neuron in the teacher. Through the experiments in various aspects of knowledge transfer, it is verified that the proposed method outperforms the current state-of-the-art.

Comments: Accepted to AAAI 2019
Categories: cs.LG, cs.CV, stat.ML
Related articles: Most relevant | Search more
arXiv:1801.00905 [cs.LG] (Published 2018-01-03)
Neural Networks in Adversarial Setting and Ill-Conditioned Weight Space
arXiv:1806.01547 [cs.LG] (Published 2018-06-05)
ClusterNet : Semi-Supervised Clustering using Neural Networks
arXiv:1805.09370 [cs.LG] (Published 2018-05-23)
Towards Robust Training of Neural Networks by Regularizing Adversarial Gradients