Computer Science > Machine Learning
[Submitted on 8 Feb 2020 (v1), last revised 22 Jun 2020 (this version, v9)]
Title:Soft Threshold Weight Reparameterization for Learnable Sparsity
View PDFAbstract:Sparsity in Deep Neural Networks (DNNs) is studied extensively with the focus of maximizing prediction accuracy given an overall parameter budget. Existing methods rely on uniform or heuristic non-uniform sparsity budgets which have sub-optimal layer-wise parameter allocation resulting in a) lower prediction accuracy or b) higher inference cost (FLOPs). This work proposes Soft Threshold Reparameterization (STR), a novel use of the soft-threshold operator on DNN weights. STR smoothly induces sparsity while learning pruning thresholds thereby obtaining a non-uniform sparsity budget. Our method achieves state-of-the-art accuracy for unstructured sparsity in CNNs (ResNet50 and MobileNetV1 on ImageNet-1K), and, additionally, learns non-uniform budgets that empirically reduce the FLOPs by up to 50%. Notably, STR boosts the accuracy over existing results by up to 10% in the ultra sparse (99%) regime and can also be used to induce low-rank (structured sparsity) in RNNs. In short, STR is a simple mechanism which learns effective sparsity budgets that contrast with popular heuristics. Code, pretrained models and sparsity budgets are at this https URL.
Submission history
From: Aditya Kusupati [view email][v1] Sat, 8 Feb 2020 21:31:25 UTC (380 KB)
[v2] Fri, 14 Feb 2020 22:57:06 UTC (323 KB)
[v3] Fri, 21 Feb 2020 10:11:37 UTC (324 KB)
[v4] Wed, 11 Mar 2020 09:57:20 UTC (324 KB)
[v5] Fri, 17 Apr 2020 12:57:04 UTC (326 KB)
[v6] Wed, 22 Apr 2020 02:16:28 UTC (267 KB)
[v7] Tue, 28 Apr 2020 01:12:37 UTC (267 KB)
[v8] Sun, 10 May 2020 02:39:39 UTC (267 KB)
[v9] Mon, 22 Jun 2020 23:37:12 UTC (1,694 KB)
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Connected Papers (What is Connected Papers?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
alphaXiv (What is alphaXiv?)
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Hugging Face (What is Huggingface?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
CORE Recommender (What is CORE?)
IArxiv Recommender
(What is IArxiv?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.