Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox
Picture for Xiaolong Ma

Xiaolong Ma

MEST: Accurate and Fast Memory-Economic Sparse Training Framework on the Edge


Oct 26, 2021
Geng Yuan, Xiaolong Ma, Wei Niu, Zhengang Li, Zhenglun Kong, Ning Liu, Yifan Gong, Zheng Zhan, Chaoyang He, Qing Jin, Siyue Wang, Minghai Qin, Bin Ren, Yanzhi Wang, Sijia Liu, Xue Lin

Figure 1 for MEST: Accurate and Fast Memory-Economic Sparse Training Framework on the Edge
Figure 2 for MEST: Accurate and Fast Memory-Economic Sparse Training Framework on the Edge
Figure 3 for MEST: Accurate and Fast Memory-Economic Sparse Training Framework on the Edge
Figure 4 for MEST: Accurate and Fast Memory-Economic Sparse Training Framework on the Edge
* NeurIPS 2021 Spotlight Paper 
  Access Paper or Ask Questions

GRIM: A General, Real-Time Deep Learning Inference Framework for Mobile Devices based on Fine-Grained Structured Weight Sparsity


Aug 25, 2021
Wei Niu, Zhengang Li, Xiaolong Ma, Peiyan Dong, Gang Zhou, Xuehai Qian, Xue Lin, Yanzhi Wang, Bin Ren

Figure 1 for GRIM: A General, Real-Time Deep Learning Inference Framework for Mobile Devices based on Fine-Grained Structured Weight Sparsity
Figure 2 for GRIM: A General, Real-Time Deep Learning Inference Framework for Mobile Devices based on Fine-Grained Structured Weight Sparsity
Figure 3 for GRIM: A General, Real-Time Deep Learning Inference Framework for Mobile Devices based on Fine-Grained Structured Weight Sparsity
Figure 4 for GRIM: A General, Real-Time Deep Learning Inference Framework for Mobile Devices based on Fine-Grained Structured Weight Sparsity
* Accepted in IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2021 
  Access Paper or Ask Questions

A Topological-Framework to Improve Analysis of Machine Learning Model Performance


Jul 09, 2021
Henry Kvinge, Colby Wight, Sarah Akers, Scott Howland, Woongjo Choi, Xiaolong Ma, Luke Gosink, Elizabeth Jurrus, Keerti Kappagantula, Tegan H. Emerson

Figure 1 for A Topological-Framework to Improve Analysis of Machine Learning Model Performance
Figure 2 for A Topological-Framework to Improve Analysis of Machine Learning Model Performance
* 6 pages 
  Access Paper or Ask Questions

Sanity Checks for Lottery Tickets: Does Your Winning Ticket Really Win the Jackpot?


Jul 01, 2021
Xiaolong Ma, Geng Yuan, Xuan Shen, Tianlong Chen, Xuxi Chen, Xiaohan Chen, Ning Liu, Minghai Qin, Sijia Liu, Zhangyang Wang, Yanzhi Wang

Figure 1 for Sanity Checks for Lottery Tickets: Does Your Winning Ticket Really Win the Jackpot?
Figure 2 for Sanity Checks for Lottery Tickets: Does Your Winning Ticket Really Win the Jackpot?
Figure 3 for Sanity Checks for Lottery Tickets: Does Your Winning Ticket Really Win the Jackpot?
Figure 4 for Sanity Checks for Lottery Tickets: Does Your Winning Ticket Really Win the Jackpot?

  Access Paper or Ask Questions

Improving DNN Fault Tolerance using Weight Pruning and Differential Crossbar Mapping for ReRAM-based Edge AI


Jun 18, 2021
Geng Yuan, Zhiheng Liao, Xiaolong Ma, Yuxuan Cai, Zhenglun Kong, Xuan Shen, Jingyan Fu, Zhengang Li, Chengming Zhang, Hongwu Peng, Ning Liu, Ao Ren, Jinhui Wang, Yanzhi Wang

Figure 1 for Improving DNN Fault Tolerance using Weight Pruning and Differential Crossbar Mapping for ReRAM-based Edge AI
Figure 2 for Improving DNN Fault Tolerance using Weight Pruning and Differential Crossbar Mapping for ReRAM-based Edge AI
Figure 3 for Improving DNN Fault Tolerance using Weight Pruning and Differential Crossbar Mapping for ReRAM-based Edge AI
Figure 4 for Improving DNN Fault Tolerance using Weight Pruning and Differential Crossbar Mapping for ReRAM-based Edge AI
* In Proceedings of the 22nd International Symposium on Quality Electronic Design (ISQED), 2021 
  Access Paper or Ask Questions

Effective Model Sparsification by Scheduled Grow-and-Prune Methods


Jun 18, 2021
Xiaolong Ma, Minghai Qin, Fei Sun, Zejiang Hou, Kun Yuan, Yi Xu, Yanzhi Wang, Yen-Kuang Chen, Rong Jin, Yuan Xie

Figure 1 for Effective Model Sparsification by Scheduled Grow-and-Prune Methods
Figure 2 for Effective Model Sparsification by Scheduled Grow-and-Prune Methods
Figure 3 for Effective Model Sparsification by Scheduled Grow-and-Prune Methods
Figure 4 for Effective Model Sparsification by Scheduled Grow-and-Prune Methods

  Access Paper or Ask Questions

FORMS: Fine-grained Polarized ReRAM-based In-situ Computation for Mixed-signal DNN Accelerator


Jun 16, 2021
Geng Yuan, Payman Behnam, Zhengang Li, Ali Shafiee, Sheng Lin, Xiaolong Ma, Hang Liu, Xuehai Qian, Mahdi Nazm Bojnordi, Yanzhi Wang, Caiwen Ding

Figure 1 for FORMS: Fine-grained Polarized ReRAM-based In-situ Computation for Mixed-signal DNN Accelerator
Figure 2 for FORMS: Fine-grained Polarized ReRAM-based In-situ Computation for Mixed-signal DNN Accelerator
Figure 3 for FORMS: Fine-grained Polarized ReRAM-based In-situ Computation for Mixed-signal DNN Accelerator
Figure 4 for FORMS: Fine-grained Polarized ReRAM-based In-situ Computation for Mixed-signal DNN Accelerator
* In Proceedings of the 48th Annual International Symposium on Computer Architecture (ISCA), 2021 
  Access Paper or Ask Questions

Towards Fast and Accurate Multi-Person Pose Estimation on Mobile Devices


Jun 06, 2021
Xuan Shen, Geng Yuan, Wei Niu, Xiaolong Ma, Jiexiong Guan, Zhengang Li, Bin Ren, Yanzhi Wang

Figure 1 for Towards Fast and Accurate Multi-Person Pose Estimation on Mobile Devices
Figure 2 for Towards Fast and Accurate Multi-Person Pose Estimation on Mobile Devices
Figure 3 for Towards Fast and Accurate Multi-Person Pose Estimation on Mobile Devices
Figure 4 for Towards Fast and Accurate Multi-Person Pose Estimation on Mobile Devices

  Access Paper or Ask Questions

Lottery Ticket Implies Accuracy Degradation, Is It a Desirable Phenomenon?


Feb 19, 2021
Ning Liu, Geng Yuan, Zhengping Che, Xuan Shen, Xiaolong Ma, Qing Jin, Jian Ren, Jian Tang, Sijia Liu, Yanzhi Wang

Figure 1 for Lottery Ticket Implies Accuracy Degradation, Is It a Desirable Phenomenon?
Figure 2 for Lottery Ticket Implies Accuracy Degradation, Is It a Desirable Phenomenon?
Figure 3 for Lottery Ticket Implies Accuracy Degradation, Is It a Desirable Phenomenon?
Figure 4 for Lottery Ticket Implies Accuracy Degradation, Is It a Desirable Phenomenon?

  Access Paper or Ask Questions

A Unified DNN Weight Compression Framework Using Reweighted Optimization Methods


Apr 12, 2020
Tianyun Zhang, Xiaolong Ma, Zheng Zhan, Shanglin Zhou, Minghai Qin, Fei Sun, Yen-Kuang Chen, Caiwen Ding, Makan Fardad, Yanzhi Wang

Figure 1 for A Unified DNN Weight Compression Framework Using Reweighted Optimization Methods
Figure 2 for A Unified DNN Weight Compression Framework Using Reweighted Optimization Methods
Figure 3 for A Unified DNN Weight Compression Framework Using Reweighted Optimization Methods
Figure 4 for A Unified DNN Weight Compression Framework Using Reweighted Optimization Methods

  Access Paper or Ask Questions

A Privacy-Preserving DNN Pruning and Mobile Acceleration Framework


Mar 13, 2020
Zheng Zhan, Yifan Gong, Zhengang Li, Pu Zhao, Xiaolong Ma, Wei Niu, Xiaolin Xu, Bin Ren, Yanzhi Wang, Xue Lin

Figure 1 for A Privacy-Preserving DNN Pruning and Mobile Acceleration Framework
Figure 2 for A Privacy-Preserving DNN Pruning and Mobile Acceleration Framework
Figure 3 for A Privacy-Preserving DNN Pruning and Mobile Acceleration Framework
Figure 4 for A Privacy-Preserving DNN Pruning and Mobile Acceleration Framework

  Access Paper or Ask Questions

BLK-REW: A Unified Block-based DNN Pruning Framework using Reweighted Regularization Method


Feb 22, 2020
Xiaolong Ma, Zhengang Li, Yifan Gong, Tianyun Zhang, Wei Niu, Zheng Zhan, Pu Zhao, Jian Tang, Xue Lin, Bin Ren, Yanzhi Wang

Figure 1 for BLK-REW: A Unified Block-based DNN Pruning Framework using Reweighted Regularization Method
Figure 2 for BLK-REW: A Unified Block-based DNN Pruning Framework using Reweighted Regularization Method
Figure 3 for BLK-REW: A Unified Block-based DNN Pruning Framework using Reweighted Regularization Method
Figure 4 for BLK-REW: A Unified Block-based DNN Pruning Framework using Reweighted Regularization Method

  Access Paper or Ask Questions

An Image Enhancing Pattern-based Sparsity for Real-time Inference on Mobile Devices


Feb 22, 2020
Xiaolong Ma, Wei Niu, Tianyun Zhang, Sijia Liu, Sheng Lin, Hongjia Li, Xiang Chen, Jian Tang, Kaisheng Ma, Bin Ren, Yanzhi Wang

Figure 1 for An Image Enhancing Pattern-based Sparsity for Real-time Inference on Mobile Devices
Figure 2 for An Image Enhancing Pattern-based Sparsity for Real-time Inference on Mobile Devices
Figure 3 for An Image Enhancing Pattern-based Sparsity for Real-time Inference on Mobile Devices
Figure 4 for An Image Enhancing Pattern-based Sparsity for Real-time Inference on Mobile Devices
* arXiv admin note: text overlap with arXiv:1909.05073 
  Access Paper or Ask Questions

PCNN: Pattern-based Fine-Grained Regular Pruning towards Optimizing CNN Accelerators


Feb 11, 2020
Zhanhong Tan, Jiebo Song, Xiaolong Ma, Sia-Huat Tan, Hongyang Chen, Yuanqing Miao, Yifu Wu, Shaokai Ye, Yanzhi Wang, Dehui Li, Kaisheng Ma

Figure 1 for PCNN: Pattern-based Fine-Grained Regular Pruning towards Optimizing CNN Accelerators
Figure 2 for PCNN: Pattern-based Fine-Grained Regular Pruning towards Optimizing CNN Accelerators
Figure 3 for PCNN: Pattern-based Fine-Grained Regular Pruning towards Optimizing CNN Accelerators
Figure 4 for PCNN: Pattern-based Fine-Grained Regular Pruning towards Optimizing CNN Accelerators
* 6 pages, DAC 2020 accepted paper 
  Access Paper or Ask Questions

SS-Auto: A Single-Shot, Automatic Structured Weight Pruning Framework of DNNs with Ultra-High Efficiency


Jan 23, 2020
Zhengang Li, Yifan Gong, Xiaolong Ma, Sijia Liu, Mengshu Sun, Zheng Zhan, Zhenglun Kong, Geng Yuan, Yanzhi Wang

Figure 1 for SS-Auto: A Single-Shot, Automatic Structured Weight Pruning Framework of DNNs with Ultra-High Efficiency
Figure 2 for SS-Auto: A Single-Shot, Automatic Structured Weight Pruning Framework of DNNs with Ultra-High Efficiency
Figure 3 for SS-Auto: A Single-Shot, Automatic Structured Weight Pruning Framework of DNNs with Ultra-High Efficiency
Figure 4 for SS-Auto: A Single-Shot, Automatic Structured Weight Pruning Framework of DNNs with Ultra-High Efficiency

  Access Paper or Ask Questions

PatDNN: Achieving Real-Time DNN Execution on Mobile Devices with Pattern-based Weight Pruning


Jan 22, 2020
Wei Niu, Xiaolong Ma, Sheng Lin, Shihao Wang, Xuehai Qian, Xue Lin, Yanzhi Wang, Bin Ren

Figure 1 for PatDNN: Achieving Real-Time DNN Execution on Mobile Devices with Pattern-based Weight Pruning
Figure 2 for PatDNN: Achieving Real-Time DNN Execution on Mobile Devices with Pattern-based Weight Pruning
Figure 3 for PatDNN: Achieving Real-Time DNN Execution on Mobile Devices with Pattern-based Weight Pruning
Figure 4 for PatDNN: Achieving Real-Time DNN Execution on Mobile Devices with Pattern-based Weight Pruning
* To be published in the Proceedings of Twenty-Fifth International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS 20) 
  Access Paper or Ask Questions

A SOT-MRAM-based Processing-In-Memory Engine for Highly Compressed DNN Implementation


Nov 24, 2019
Geng Yuan, Xiaolong Ma, Sheng Lin, Zhengang Li, Caiwen Ding

Figure 1 for A SOT-MRAM-based Processing-In-Memory Engine for Highly Compressed DNN Implementation
Figure 2 for A SOT-MRAM-based Processing-In-Memory Engine for Highly Compressed DNN Implementation
Figure 3 for A SOT-MRAM-based Processing-In-Memory Engine for Highly Compressed DNN Implementation
Figure 4 for A SOT-MRAM-based Processing-In-Memory Engine for Highly Compressed DNN Implementation

  Access Paper or Ask Questions

PCONV: The Missing but Desirable Sparsity in DNN Weight Pruning for Real-time Execution on Mobile Devices


Sep 12, 2019
Xiaolong Ma, Fu-Ming Guo, Wei Niu, Xue Lin, Jian Tang, Kaisheng Ma, Bin Ren, Yanzhi Wang

Figure 1 for PCONV: The Missing but Desirable Sparsity in DNN Weight Pruning for Real-time Execution on Mobile Devices
Figure 2 for PCONV: The Missing but Desirable Sparsity in DNN Weight Pruning for Real-time Execution on Mobile Devices
Figure 3 for PCONV: The Missing but Desirable Sparsity in DNN Weight Pruning for Real-time Execution on Mobile Devices
Figure 4 for PCONV: The Missing but Desirable Sparsity in DNN Weight Pruning for Real-time Execution on Mobile Devices

  Access Paper or Ask Questions

An Ultra-Efficient Memristor-Based DNN Framework with Structured Weight Pruning and Quantization Using ADMM


Aug 29, 2019
Geng Yuan, Xiaolong Ma, Caiwen Ding, Sheng Lin, Tianyun Zhang, Zeinab S. Jalali, Yilong Zhao, Li Jiang, Sucheta Soundarajan, Yanzhi Wang

Figure 1 for An Ultra-Efficient Memristor-Based DNN Framework with Structured Weight Pruning and Quantization Using ADMM
Figure 2 for An Ultra-Efficient Memristor-Based DNN Framework with Structured Weight Pruning and Quantization Using ADMM
Figure 3 for An Ultra-Efficient Memristor-Based DNN Framework with Structured Weight Pruning and Quantization Using ADMM
Figure 4 for An Ultra-Efficient Memristor-Based DNN Framework with Structured Weight Pruning and Quantization Using ADMM

  Access Paper or Ask Questions

Tiny but Accurate: A Pruned, Quantized and Optimized Memristor Crossbar Framework for Ultra Efficient DNN Implementation


Aug 27, 2019
Xiaolong Ma, Geng Yuan, Sheng Lin, Caiwen Ding, Fuxun Yu, Tao Liu, Wujie Wen, Xiang Chen, Yanzhi Wang

Figure 1 for Tiny but Accurate: A Pruned, Quantized and Optimized Memristor Crossbar Framework for Ultra Efficient DNN Implementation
Figure 2 for Tiny but Accurate: A Pruned, Quantized and Optimized Memristor Crossbar Framework for Ultra Efficient DNN Implementation
Figure 3 for Tiny but Accurate: A Pruned, Quantized and Optimized Memristor Crossbar Framework for Ultra Efficient DNN Implementation
Figure 4 for Tiny but Accurate: A Pruned, Quantized and Optimized Memristor Crossbar Framework for Ultra Efficient DNN Implementation

  Access Paper or Ask Questions

AutoSlim: An Automatic DNN Structured Pruning Framework for Ultra-High Compression Rates


Jul 06, 2019
Ning Liu, Xiaolong Ma, Zhiyuan Xu, Yanzhi Wang, Jian Tang, Jieping Ye

Figure 1 for AutoSlim: An Automatic DNN Structured Pruning Framework for Ultra-High Compression Rates
Figure 2 for AutoSlim: An Automatic DNN Structured Pruning Framework for Ultra-High Compression Rates
Figure 3 for AutoSlim: An Automatic DNN Structured Pruning Framework for Ultra-High Compression Rates
Figure 4 for AutoSlim: An Automatic DNN Structured Pruning Framework for Ultra-High Compression Rates

  Access Paper or Ask Questions

Non-structured DNN Weight Pruning Considered Harmful


Jul 03, 2019
Yanzhi Wang, Shaokai Ye, Zhezhi He, Xiaolong Ma, Linfeng Zhang, Sheng Lin, Geng Yuan, Sia Huat Tan, Zhengang Li, Deliang Fan, Xuehai Qian, Xue Lin, Kaisheng Ma

Figure 1 for Non-structured DNN Weight Pruning Considered Harmful
Figure 2 for Non-structured DNN Weight Pruning Considered Harmful
Figure 3 for Non-structured DNN Weight Pruning Considered Harmful
Figure 4 for Non-structured DNN Weight Pruning Considered Harmful

  Access Paper or Ask Questions

Toward Extremely Low Bit and Lossless Accuracy in DNNs with Progressive ADMM


May 02, 2019
Sheng Lin, Xiaolong Ma, Shaokai Ye, Geng Yuan, Kaisheng Ma, Yanzhi Wang

Figure 1 for Toward Extremely Low Bit and Lossless Accuracy in DNNs with Progressive ADMM
Figure 2 for Toward Extremely Low Bit and Lossless Accuracy in DNNs with Progressive ADMM
Figure 3 for Toward Extremely Low Bit and Lossless Accuracy in DNNs with Progressive ADMM
Figure 4 for Toward Extremely Low Bit and Lossless Accuracy in DNNs with Progressive ADMM
* Accepted by ICML workshop (ODML-CDNNR2019). arXiv admin note: substantial text overlap with arXiv:1903.09769 
  Access Paper or Ask Questions

26ms Inference Time for ResNet-50: Towards Real-Time Execution of all DNNs on Smartphone


May 02, 2019
Wei Niu, Xiaolong Ma, Yanzhi Wang, Bin Ren

Figure 1 for 26ms Inference Time for ResNet-50: Towards Real-Time Execution of all DNNs on Smartphone
Figure 2 for 26ms Inference Time for ResNet-50: Towards Real-Time Execution of all DNNs on Smartphone
Figure 3 for 26ms Inference Time for ResNet-50: Towards Real-Time Execution of all DNNs on Smartphone
Figure 4 for 26ms Inference Time for ResNet-50: Towards Real-Time Execution of all DNNs on Smartphone

  Access Paper or Ask Questions

ResNet Can Be Pruned 60x: Introducing Network Purification and Unused Path Removal (P-RM) after Weight Pruning


Apr 30, 2019
Xiaolong Ma, Geng Yuan, Sheng Lin, Zhengang Li, Hao Sun, Yanzhi Wang

Figure 1 for ResNet Can Be Pruned 60x: Introducing Network Purification and Unused Path Removal (P-RM) after Weight Pruning
Figure 2 for ResNet Can Be Pruned 60x: Introducing Network Purification and Unused Path Removal (P-RM) after Weight Pruning
Figure 3 for ResNet Can Be Pruned 60x: Introducing Network Purification and Unused Path Removal (P-RM) after Weight Pruning
Figure 4 for ResNet Can Be Pruned 60x: Introducing Network Purification and Unused Path Removal (P-RM) after Weight Pruning
* Submitted to ICML workshop 
  Access Paper or Ask Questions