AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

Balanced Sparsity for Efficient DNN Inference on GPU

Zhuliang Yao, Shijie Cao, Wencong Xiao, Chen Zhang, Lanshun Nie

AAAI 2019 · Architectures and efficiency › Pruning and efficient tuning · approach: Sparse training from scratch · 93 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace random sparsity with Balanced SparsityPTB LSTM inference · PTBinference time (us): 370.9 -> 120.2Improved
replace dense computation with Balanced SparsityPTB LSTM inference · PTBinference time (us): 294.1 -> 120.2Improved
replace block sparsity with Balanced SparsityPTB LSTM inference · PTBinference time (us): 326.3 -> 120.2Improved
replace random sparsity with Balanced SparsityTIMIT CTC Bi-LSTM inference · TIMITinference time (us): 190.5 -> 83.9Improved
replace dense computation with Balanced SparsityTIMIT CTC Bi-LSTM inference · TIMITinference time (us): 117.9 -> 83.9Improved
replace block sparsity with Balanced SparsityTIMIT CTC Bi-LSTM inference · TIMITinference time (us): 212.8 -> 83.9Improved
replace random sparsity with Balanced SparsityVGG-16 inference · ImageNet ILSVRC-2012inference time (us): 33407.4 -> 5213.0Improved
use Balanced Sparsity pruningVGG-16 · ImageNet ILSVRC-2012top-5 accuracy: 90.3%No clear effect
use balance range 25 instead of other tested ranges (ablation)Balanced Sparsity on PTB LSTM · PTBperplexity: 78.3; 78.4; 78.4No clear effect
use block size 16*16 rather than 4*4 (ablation)Block Sparsity on PTB LSTM · PTB, 80% sparsityperplexity: 88.1 -> 99.5Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy