Skip to the content.

Research Assistant (2017-current)

Department of Computer Science

Emory University

Email: jwan936@emory.edu

Google Scholar CV

News

2022/6: I serve as a PC member for the EMNLP 2022 and an independent reviewer for the Knowledge-based System journal.

2022/5: One paper is accepted by KDD 2022.

2022/5: I serve as a PC member for the NeurIPS 2022 and ACML 2022.

2022/4: I serve as a PC member for the ECML-PKDD 2022 and an independent reviwer for the Expert Systems With Applications journal.

2022/2: I serve as a PC member for the IJCNN 2022 and an independent reviewer for the TKDE journal.

2022/2: One paper is accepted by Neurocomputing. I have published ten first-author papers during my Ph.D.

2022/2: I am invited to give a talk at the EURO 2022.

2022/1: One paper is accepted by WWW 2022.

2022/1: One paper is accepted by PAKDD 2022.

2021/12: I am invited to give a talk at the INFORMS Optimization Society Conference 2022.

2021/12: The source code of the dlADMM algorithm for the GCN model has been released. Feel free to contact me if you have any questions.

2021/12: I serve as a PC member for the IJCAI 2022.

2021/10: I serve as an independent reviewer for the journal Applied Artifical Intelligence.

2021/10: I serve as an independent reviewer for the journal Numerical Algorithms (SJR Q1).

2021/8: I am invited to give a talk at the MOPTA 2021 conference. See you at Lehigh University!

2021/3: One paper is accepted by the Results in Control and Optimization.

Biography

Junxiang Wang is a highly motived Ph.D. student at Emory University, supervised by Dr. Liang Zhao. He has broad interest in different research areas such as social media mining, inverse problems and nonconvex optimization. Specifically, he is passionate about applications of Alterating Direction Method of Multipliers in deep learning models. He has published various research papers in top-tier conferences such as KDD, ICDM and WWW . He has been selected as an independent reviewer in many top-tiers venues in AI and operation research fields such as IJCAI and European Journal of Operational Research (EJOR). He has been inivited to present his work in many optimization conferences. Before joining Emory, he received his master degree and bachelor degree from George Mason University in 2020 and East China Normal University in 2012, respectively.

Publications

Conference Paper

(9) [KDD 2022]. Chen Ling, Junji Jiang, Junxiang Wang and Liang Zhao. SL-VAE: Variational Autoencoder for Source Localization. in Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD 2022), research track (acceptance rate: 15.0%), Washington D.C, USA, Aug 2022.

(8) [WWW 2022]. Junxiang Wang, Junji Jiang, and Liang Zhao. An Invertible Graph Diffusion Neural Network for Source Localization. 31th International World Wide Web Conference (WWW 2022), (acceptance rate: 17.7%), Lyon, FR, Apr 2022. paper code slides

(7) [PAKDD 2022]. Junxiang Wang and Liang Zhao. Convergence and Applications of ADMM on the Multi-convex Problems. 26th Pacific-Asia Conference on Knowledge Discovery and Data Mining (PAKDD 2022), (acceptance rate: 19.3%), Chengdu, China, May 2022. paper code slides

(6) [ICDM 2020]. Junxiang Wang, Zheng Chai, Yue Cheng, Liang Zhao. Toward Model Parallelism for Deep Neural Network based on Gradient-free ADMM Framework. in Proceedings of the IEEE International Conference on Data Mining (ICDM 2020), regular paper (acceptance rate: 9.8%), Sorrento, Italy, Nov 2020. paper code slides

(5) [KDD 2019]. Junxiang Wang, Fuxun Yu, Xiang Chen, and Liang Zhao. ADMM for Efficient Deep Learning with Global Convergence. in Proceedings of the 25th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD 2019), research track (acceptance rate: 14.2%), Alaska, USA, Aug 2019. paper material code video slides poster summary Chinese media coverage

(4) [BigData 2018]. Junxiang Wang, Liang Zhao, and Yanfang Ye. Semi-supervised Multi-instance Interpretable Models for Flu Shot Adverse Event Detection. 2018 IEEE International Conference on Big Data (BigData 2018) (acceptance rate: 18.9%), Seattle, USA, Dec 2018. paper code slides

(3) [ICDM 2018]. Junxiang Wang, Yuyang Gao, Andreas Zufle, Jingyuan Yang, and Liang Zhao. Incomplete Label Uncertainty Estimation for Petition Victory Prediction with Dynamic Features. in Proceedings of the IEEE International Conference on Data Mining (ICDM 2018), regular paper (acceptance rate: 8.9%), Singapore, Dec 2018. paper code data slides poster

(2) [WWW 2018]. Junxiang Wang and Liang Zhao. Multi-instance Domain Adaptation for Vaccine Adverse Event Detection. 27th International World Wide Web Conference (WWW 2018), (acceptance rate: 14.8%), Lyon, FR, Apr 2018. paper code slides

(1) [AAAI 2018]. Liang Zhao, Junxiang Wang, and Xiaojie Guo. Distant-supervision of heterogeneous multitask learning for social event forecasting with multilingual indicators. Thirty-Second AAAI Conference on Artificial Intelligence (AAAI 2018), Oral presentation (acceptance rate: 11.0%), pp. 4498-4505, New Orleans, US, Feb 2018. paper

Journal Paper

(9) [Neurocomputing]. Junxiang Wang, Hongyi Li, and Liang Zhao. Accelerated Gradient-free Neural Network Training by Multi-convex Alternating Optimization. Neurocomputing, (impact factor: 5.719), 2022. paper code

(8) [RICO]. Junxiang Wang and Liang Zhao. Nonconvex Generalization of Alternating Direction Method of Multipliers for Nonlinear Equality Constrained Problems. Results in Control and Optimization, 2021. paper code

(7) [J. Biomed. Semant]. Junxiang Wang, Liang Zhao, Yanfang Ye, and Yuji Zhang. Adverse event detection by integrating Twitter data and VAERS. Journal of Biomedical Semantics, (impact factor: 1.845), 2018. paper

(6) [Proc. IEEE]. Liang Zhao, Junxiang Wang, Feng Chen, Chang-Tien Lu and Naren Ramakrishnan. “Spatial Event Forecasting in Social Media with Geographically Hierarchical Regularization”. Proceedings of the IEEE (impact factor: 9.237), vol. 105, no. 10, pp. 1953-1970, Oct. 2017. paper

(5) [Lett.Drug.Des.Discov]. Junxiang Wang, Weiming Yu, Zhibin Chen, Hengda Li, Zhenran Jiang. Predicting Drug-Target Interactions of Nuclear Receptors Based on Molecular Descriptors Information. Letters in Drug Design & Discovery 10 (10), 989-994, 2013. paper

(4) [Pharmacogenomics]. Weiming Yu, Yan Yan, Qing Liu, Junxiang Wang and Zhenran Jiang. Predicting drug–target interaction networks of human diseases based on multiple feature information. Pharmacogenomics 14 (14), 1701-1707, 20, 2013. paper

(3) [Curr. Bioinform]. Zhenran Jiang, Ran Tao, Lei Du, Weiming Yu and Junxiang Wang. Using Network-Based Approaches to Predict Ligands of Orphan Nuclear Receptors. Current Bioinformatics 7 (4), 411-414, 2012. paper

(2) [Comb. Chem. High Throughput Screen]. Ran Tao, Zhenran Jiang, Weiming Yu and Junxiang Wang. Predicting Coupling Specificity of GPCRs Based on the Optimization of the Coupling Regions. Combinatorial chemistry & high throughput screening 15 (9), 770-774, 2012. paper

(1) [Curr. Med. Chem]. Weiming Yu, Zhengyan Jiang, Junxiang Wang and Ran Tao. Using feature selection technique for drug-target interaction networks prediction. Current medicinal chemistry 18 (36), 5687-5693, 2011. paper

Workshop Paper

(4) [OPT 2021]. Hongyi Li, Junxiang Wang, Yongchao Wang, Yue Cheng and Liang Zhao. Community-based Layerwise Distributed Training of Graph Convolutional Networks. NeurIPS 2021 Workshop on Optimization for Machine Learning (OPT 2021). paper poster

(3) [ICML WS 2021]. Junxiang Wang, Hongyi Li, Yongchao Wang and Liang Zhao. Accelerated Gradient-free Neural Network Training by Multi-convex Alternating Optimization. ICML 2021 Workshop on Beyond First-Order Methods in ML systems. paper slides video

(2) [ICML WS 2020]. Junxiang Wang, Zheng Chai, Yue Cheng, Liang Zhao. Tunable Subnetwork Splitting for Model-parallelism of Neural Network Training. ICML 2020 Workshop on Beyond First-Order Methods in ML systems. paper code video slides

(1) [OPT 2019]. Junxiang Wang and Liang Zhao. The Application of Multi-block ADMM on Isotonic Regression Problems. 11th Workshop on Optimization for Machine Learning (OPT 2019), co-located with NeurIPS 2019. paper code poster

Priprint Paper

Click to expand! **Junxiang Wang**, Hongyi Li, and Liang Zhao. Proximal ADMM Algorithms for Multi-convex Problems. **Junxiang Wang**, Junji Jiang, and Liang Zhao. An Invertible Bi-Lipschitz Surrogate Model for Black-box Graph Inverse Problems. Hongyi Li, **Junxiang Wang**, Yongchao Wang, Yue Cheng and Liang Zhao. Community-based Layerwise Distributed Training of Graph Convolutional Networks. Chen Ling, Junji Jiang, **Junxiang Wang**, Renhao Xue and Zhao Liang. DeepIM: Deep Graph Representation Learning and Optimization for Influence Maximization. **Junxiang Wang**, Hongyi Li, Liang Zhao. A Convergent ADMM Framework for Efficient Neural Network Training. **Junxiang Wang**, Xuchao Zhang, Bo Zong, Yanchi Liu, Wei Cheng, Jingchao Ni, Haifeng Chen, and Liang Zhao. Do Multi-Lingual Pre-trained Language Models Reveal Consistent Token Attributions in Different Languages? [paper](https://arxiv.org/pdf/2112.12356.pdf) Yuyang Gao,**Junxiang Wang**, Wei Wang, Xin Deng, Hamed Zamani, Xiaohan Yan, Yan Guo, Ahmed Awadallah, Yanfang Ye, and Liang Zhao. Asynchronous Semi-supervised Representation Learning for Email Heterogeneous Networks. **Junxiang Wang**, Hongyi Li (first-coauthor), Zheng Chai, Yongchao Wang, Yue Cheng, and Liang Zhao. Towards Quantized Model Parallelism for Graph-Augmented MLPs Based on Gradient-Free ADMM Framework. Preprint. [paper](https://arxiv.org/pdf/2105.09837.pdf) Johnny Torres, Guangji Bai, **Junxiang Wang**, Liang Zhao, Carmen Vaca, Cristina Abad. Sign-regularized Multi-task Learning. [paper](https://arxiv.org/pdf/2102.11191.pdf) **Junxiang Wang**, Liang Zhao, Yanfang Ye and Houman Homayoun. Interpretability Evaluation Framework for Deep Neural Networks.

Invited Talks

EURO 2022 Espoo, Finland. Invited oral presentation.

Towards Quantized Model Parallelism for Graph-Augmented MLPs Based on Gradient-Free ADMM Framework. INFORMS Optimization Society Conference (IOS) 2022. Greenville, SC. March 2022. Invited oral presentation. slides

Power of Alternating Direction Method of Multipliers(ADMM) in deep learning. Modeling and Optimization: Theory and Applications (MOPTA) 2021. Lehigh University. PA. August 2021. Invited oral presentation. slides