Prune Your Model Before Distill It
- Authors
- Park, J.; No, A.
- Issue Date
- 1-Jan-2022
- Publisher
- Springer Science and Business Media Deutschland GmbH
- Keywords
- Knowledge distillation; Label smoothing regularization (LSR); Neural network compression; Pruning
- Citation
- Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), v.13671 LNCS, pp.120 - 136
- Journal Title
- Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
- Volume
- 13671 LNCS
- Start Page
- 120
- End Page
- 136
- URI
- https://scholarworks.bwise.kr/hongik/handle/2020.sw.hongik/30622
- DOI
- 10.1007/978-3-031-20083-0_8
- ISSN
- 0302-9743
- Abstract
- Knowledge distillation transfers the knowledge from a cumbersome teacher to a small student. Recent results suggest that the student-friendly teacher is more appropriate to distill since it provides more transferrable knowledge. In this work, we propose the novel framework, “prune, then distill,” that prunes the model first to make it more transferrable and then distill it to the student. We provide several exploratory examples where the pruned teacher teaches better than the original unpruned networks. We further show theoretically that the pruned teacher plays the role of regularizer in distillation, which reduces the generalization error. Based on this result, we propose a novel neural network compression scheme where the student network is formed based on the pruned teacher and then apply the “prune, then distill” strategy. The code is available at https://github.com/ososos888/prune-then-distill. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.
- Files in This Item
- There are no files associated with this item.
- Appears in
Collections - College of Engineering > School of Electronic & Electrical Engineering > 1. Journal Articles
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.