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 Guangrun Wang

Guangrun Wang

DS-Net++: Dynamic Weight Slicing for Efficient Inference in CNNs and Transformers


Sep 21, 2021
Changlin Li, Guangrun Wang, Bing Wang, Xiaodan Liang, Zhihui Li, Xiaojun Chang

Figure 1 for DS-Net++: Dynamic Weight Slicing for Efficient Inference in CNNs and Transformers
Figure 2 for DS-Net++: Dynamic Weight Slicing for Efficient Inference in CNNs and Transformers
Figure 3 for DS-Net++: Dynamic Weight Slicing for Efficient Inference in CNNs and Transformers
Figure 4 for DS-Net++: Dynamic Weight Slicing for Efficient Inference in CNNs and Transformers
* Extension of the CVPR 2021 oral paper (https://openaccess.thecvf.com/content/CVPR2021/html/Li_Dynamic_Slimmable_Network_CVPR_2021_paper.html) 
  Access Paper or Ask Questions

EfficientBERT: Progressively Searching Multilayer Perceptron via Warm-up Knowledge Distillation


Sep 16, 2021
Chenhe Dong, Guangrun Wang, Hang Xu, Jiefeng Peng, Xiaozhe Ren, Xiaodan Liang

Figure 1 for EfficientBERT: Progressively Searching Multilayer Perceptron via Warm-up Knowledge Distillation
Figure 2 for EfficientBERT: Progressively Searching Multilayer Perceptron via Warm-up Knowledge Distillation
Figure 3 for EfficientBERT: Progressively Searching Multilayer Perceptron via Warm-up Knowledge Distillation
Figure 4 for EfficientBERT: Progressively Searching Multilayer Perceptron via Warm-up Knowledge Distillation
* Findings of EMNLP 2021 
  Access Paper or Ask Questions

Pi-NAS: Improving Neural Architecture Search by Reducing Supernet Training Consistency Shift


Aug 22, 2021
Jiefeng Peng, Jiqi Zhang, Changlin Li, Guangrun Wang, Xiaodan Liang, Liang Lin

Figure 1 for Pi-NAS: Improving Neural Architecture Search by Reducing Supernet Training Consistency Shift
Figure 2 for Pi-NAS: Improving Neural Architecture Search by Reducing Supernet Training Consistency Shift
Figure 3 for Pi-NAS: Improving Neural Architecture Search by Reducing Supernet Training Consistency Shift
Figure 4 for Pi-NAS: Improving Neural Architecture Search by Reducing Supernet Training Consistency Shift
* Accepted to ICCV 2021 
  Access Paper or Ask Questions

Solving Inefficiency of Self-supervised Representation Learning


Apr 18, 2021
Guangrun Wang, Keze Wang, Guangcong Wang, Phillip H. S. Torr, Liang Lin

Figure 1 for Solving Inefficiency of Self-supervised Representation Learning
Figure 2 for Solving Inefficiency of Self-supervised Representation Learning
Figure 3 for Solving Inefficiency of Self-supervised Representation Learning
Figure 4 for Solving Inefficiency of Self-supervised Representation Learning
* 11 pages, 3 figures 
  Access Paper or Ask Questions

Joint Learning of Neural Transfer and Architecture Adaptation for Image Recognition


Mar 31, 2021
Guangrun Wang, Liang Lin, Rongcong Chen, Guangcong Wang, Jiqi Zhang

Figure 1 for Joint Learning of Neural Transfer and Architecture Adaptation for Image Recognition
Figure 2 for Joint Learning of Neural Transfer and Architecture Adaptation for Image Recognition
Figure 3 for Joint Learning of Neural Transfer and Architecture Adaptation for Image Recognition
Figure 4 for Joint Learning of Neural Transfer and Architecture Adaptation for Image Recognition
* To appear in IEEE Transactions ON Neural Networks and Learning Systems. We prove that dynamically adapting network architectures tailored for each domain task along with weight finetuning benefits in both efficiency and effectiveness, compared to the existing image recognition pipeline that only tunes the weights regardless of the architecture 
  Access Paper or Ask Questions

BossNAS: Exploring Hybrid CNN-transformers with Block-wisely Self-supervised Neural Architecture Search


Mar 24, 2021
Changlin Li, Tao Tang, Guangrun Wang, Jiefeng Peng, Bing Wang, Xiaodan Liang, Xiaojun Chang

Figure 1 for BossNAS: Exploring Hybrid CNN-transformers with Block-wisely Self-supervised Neural Architecture Search
Figure 2 for BossNAS: Exploring Hybrid CNN-transformers with Block-wisely Self-supervised Neural Architecture Search
Figure 3 for BossNAS: Exploring Hybrid CNN-transformers with Block-wisely Self-supervised Neural Architecture Search
Figure 4 for BossNAS: Exploring Hybrid CNN-transformers with Block-wisely Self-supervised Neural Architecture Search
* Our searched hybrid CNN-transformer models achieve up to 82.2% accuracy on ImageNet, surpassing EfficientNet by 2.1% with comparable compute time 
  Access Paper or Ask Questions

Dynamic Slimmable Network


Mar 24, 2021
Changlin Li, Guangrun Wang, Bing Wang, Xiaodan Liang, Zhihui Li, Xiaojun Chang

Figure 1 for Dynamic Slimmable Network
Figure 2 for Dynamic Slimmable Network
Figure 3 for Dynamic Slimmable Network
Figure 4 for Dynamic Slimmable Network
* Accepted to CVPR 2021 as an Oral Presentation 
  Access Paper or Ask Questions

EagleEye: Fast Sub-net Evaluation for Efficient Neural Network Pruning


Jul 06, 2020
Bailin Li, Bowen Wu, Jiang Su, Guangrun Wang, Liang Lin

Figure 1 for EagleEye: Fast Sub-net Evaluation for Efficient Neural Network Pruning
Figure 2 for EagleEye: Fast Sub-net Evaluation for Efficient Neural Network Pruning
Figure 3 for EagleEye: Fast Sub-net Evaluation for Efficient Neural Network Pruning
Figure 4 for EagleEye: Fast Sub-net Evaluation for Efficient Neural Network Pruning
* Accepted in ECCV 2020(Oral). Codes are available on https://github.com/anonymous47823493/EagleEye 
  Access Paper or Ask Questions

Transferable, Controllable, and Inconspicuous Adversarial Attacks on Person Re-identification With Deep Mis-Ranking


Apr 08, 2020
Hongjun Wang, Guangrun Wang, Ya Li, Dongyu Zhang, Liang Lin

Figure 1 for Transferable, Controllable, and Inconspicuous Adversarial Attacks on Person Re-identification With Deep Mis-Ranking
Figure 2 for Transferable, Controllable, and Inconspicuous Adversarial Attacks on Person Re-identification With Deep Mis-Ranking
Figure 3 for Transferable, Controllable, and Inconspicuous Adversarial Attacks on Person Re-identification With Deep Mis-Ranking
Figure 4 for Transferable, Controllable, and Inconspicuous Adversarial Attacks on Person Re-identification With Deep Mis-Ranking
* Accepted in CVPR 2020. To attack ReID, we propose a learning-to-mis-rank formulation to perturb the ranking of the system output 
  Access Paper or Ask Questions

Blockwisely Supervised Neural Architecture Search with Knowledge Distillation


Nov 29, 2019
Changlin Li, Jiefeng Peng, Liuchun Yuan, Guangrun Wang, Xiaodan Liang, Liang Lin, Xiaojun Chang

Figure 1 for Blockwisely Supervised Neural Architecture Search with Knowledge Distillation
Figure 2 for Blockwisely Supervised Neural Architecture Search with Knowledge Distillation
Figure 3 for Blockwisely Supervised Neural Architecture Search with Knowledge Distillation
Figure 4 for Blockwisely Supervised Neural Architecture Search with Knowledge Distillation
* We achieve a state-of-the-art 78.4% top-1 accuracy on ImageNet in a mobile setting, which is about a 2.1% gain over EfficientNet-B0 
  Access Paper or Ask Questions

Learnable Parameter Similarity


Jul 27, 2019
Guangcong Wang, Jianhuang Lai, Wenqi Liang, Guangrun Wang

Figure 1 for Learnable Parameter Similarity
Figure 2 for Learnable Parameter Similarity
Figure 3 for Learnable Parameter Similarity
Figure 4 for Learnable Parameter Similarity
* 9 pages 
  Access Paper or Ask Questions

Weakly Supervised Person Re-identification: Cost-effective Learning with A New Benchmark


Apr 08, 2019
Guangrun Wang, Guangcong Wang, Xujie Zhang, Jianhuang Lai, Liang Lin

Figure 1 for Weakly Supervised Person Re-identification: Cost-effective Learning with A New Benchmark
Figure 2 for Weakly Supervised Person Re-identification: Cost-effective Learning with A New Benchmark
Figure 3 for Weakly Supervised Person Re-identification: Cost-effective Learning with A New Benchmark
Figure 4 for Weakly Supervised Person Re-identification: Cost-effective Learning with A New Benchmark
* a large-scale person ReID benchmark 
  Access Paper or Ask Questions

Adaptively Connected Neural Networks


Apr 07, 2019
Guangrun Wang, Keze Wang, Liang Lin

Figure 1 for Adaptively Connected Neural Networks
Figure 2 for Adaptively Connected Neural Networks
Figure 3 for Adaptively Connected Neural Networks
Figure 4 for Adaptively Connected Neural Networks
* Accepted by CVPR 2019 
  Access Paper or Ask Questions

Deep Structured Scene Parsing by Learning with Image Descriptions


Feb 28, 2018
Liang Lin, Guangrun Wang, Rui Zhang, Ruimao Zhang, Xiaodan Liang, Wangmeng Zuo

Figure 1 for Deep Structured Scene Parsing by Learning with Image Descriptions
Figure 2 for Deep Structured Scene Parsing by Learning with Image Descriptions
Figure 3 for Deep Structured Scene Parsing by Learning with Image Descriptions
Figure 4 for Deep Structured Scene Parsing by Learning with Image Descriptions
* Discovering a semantic object hierarchy with object interaction relations (Publhised in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016. (oral)) 
  Access Paper or Ask Questions

Batch Kalman Normalization: Towards Training Deep Neural Networks with Micro-Batches


Feb 28, 2018
Guangrun Wang, Jiefeng Peng, Ping Luo, Xinjiang Wang, Liang Lin

Figure 1 for Batch Kalman Normalization: Towards Training Deep Neural Networks with Micro-Batches
Figure 2 for Batch Kalman Normalization: Towards Training Deep Neural Networks with Micro-Batches
Figure 3 for Batch Kalman Normalization: Towards Training Deep Neural Networks with Micro-Batches
Figure 4 for Batch Kalman Normalization: Towards Training Deep Neural Networks with Micro-Batches
* We presented how to improve and accelerate the training of DNNs, particularly under the context of micro-batches. (Submitted to IJCAI 2018) 
  Access Paper or Ask Questions

Hierarchical Scene Parsing by Weakly Supervised Learning with Image Descriptions


Jan 27, 2018
Ruimao Zhang, Liang Lin, Guangrun Wang, Meng Wang, Wangmeng Zuo

Figure 1 for Hierarchical Scene Parsing by Weakly Supervised Learning with Image Descriptions
Figure 2 for Hierarchical Scene Parsing by Weakly Supervised Learning with Image Descriptions
Figure 3 for Hierarchical Scene Parsing by Weakly Supervised Learning with Image Descriptions
Figure 4 for Hierarchical Scene Parsing by Weakly Supervised Learning with Image Descriptions
* Accepted by Transactions on Pattern Analysis and Machine Intelligence (T-PAMI) 2018 
  Access Paper or Ask Questions

Cross-Domain Visual Matching via Generalized Similarity Measure and Feature Learning


May 13, 2016
Liang Lin, Guangrun Wang, Wangmeng Zuo, Xiangchu Feng, Lei Zhang

Figure 1 for Cross-Domain Visual Matching via Generalized Similarity Measure and Feature Learning
Figure 2 for Cross-Domain Visual Matching via Generalized Similarity Measure and Feature Learning
Figure 3 for Cross-Domain Visual Matching via Generalized Similarity Measure and Feature Learning
Figure 4 for Cross-Domain Visual Matching via Generalized Similarity Measure and Feature Learning
* To appear in IEEE Transactions on Pattern Analysis and Machine Intelligence (T-PAMI), 2016 
  Access Paper or Ask Questions

DARI: Distance metric And Representation Integration for Person Verification


Apr 15, 2016
Guangrun Wang, Liang Lin, Shengyong Ding, Ya Li, Qing Wang

Figure 1 for DARI: Distance metric And Representation Integration for Person Verification
Figure 2 for DARI: Distance metric And Representation Integration for Person Verification
Figure 3 for DARI: Distance metric And Representation Integration for Person Verification
* To appear in Proceedings of AAAI Conference on Artificial Intelligence (AAAI), 2016 
  Access Paper or Ask Questions

Deep Feature Learning with Relative Distance Comparison for Person Re-identification


Dec 11, 2015
Shengyong Ding, Liang Lin, Guangrun Wang, Hongyang Chao

Figure 1 for Deep Feature Learning with Relative Distance Comparison for Person Re-identification
Figure 2 for Deep Feature Learning with Relative Distance Comparison for Person Re-identification
Figure 3 for Deep Feature Learning with Relative Distance Comparison for Person Re-identification
Figure 4 for Deep Feature Learning with Relative Distance Comparison for Person Re-identification
* 29 pages, 9 figures, The code has been released. http://vision.sysu.edu.cn/projects/deepreid/ 
  Access Paper or Ask Questions