Hi! I am Yifei Wang (ηδΈι£ in Chinese), a PhD candidate (expected to graduate in June 2023) at School of Mathematical Sciences, Peking University. I am a member of ZERO Lab and advised by Yisen Wang, Jiansheng Yang, and Zhouchen Lin. Previously, I received a BS in mathematics and a BA in philosophy from Peking University.
I am very interested in developing theoretical understandings and principled methods for machine learning, primarily focused on unsupervised learning (representation & generation), robust learning (adversarial & OOD robustness), and graph learning (GNN & Transformer). I am open to collaboration and please feel free to reach me by email, Wechat, or Twitter.
Some links: Email / CV / Github / Twitter / Wechat / Google Scholar
News
- 2023.02. One paper on equilibrium-based denoising accepted to TIP (first journal paper!)
- π Five papers I like very much on Contrastive learning dynamics, Non-contrastive learning dynamics, DynACL (SOTA adv robustness in SSL), ContraNorm (CL for GNN & Transformer oversmoothing), and Unbiased sampling for GNN accepted to ICLR 2023!
- One paper on AT-and-IRM connection (Oral) accepted to AAAI 2023
- Two papers on CL identifiability (Oral) & EBM-based SSL accepted to NeurIPS 2022 SSL workshop
- Three papers on MAE theory (Spotlight), Structured AT (Spotlight), and AT recipe for ViTs (Spotlight) accepted to NeurIPS 2022
- Two papers on nonlinear graph diffusion and concentrated spectral graph filters accepted to ICML 2022
- Two papers on Contrastive Learning theory and AT as EBM training accepted to ICLR 2022
- Two papers on briding invariant & equivariant SSL and GCN's continuous diffusion accepted to NeurIPS 2021
- One paper on AT as EBM training accepted to ICML 2021 AML workshop and won Silver Best Paper Award!
- One paper on reparameterized MCMC accepted to ECML-PKDD 2021 and won Best ML Paper Award (1/685)!
Publications (* marks equal contribution)
- A Message Passing Perspective on Learning Dynamics of Contrastive Learning ICLR 2023 2023 PDF
- Towards a Unified Theoretical Understanding of Non-contrastive Learning via Rank Differential Mechanism ICLR 2023 2023 PDF
- How Mask Matters: Towards Theoretical Understandings of Masked Autoencoders NeurIPS 2022 (Spotlight, Top 5%) 2022 PDF | Code | Slides
- Chaos is a Ladder: A New Theoretical Understanding of Contrastive Learning via Augmentation Overlap ICLR 2022 2022 PDF | Code | Slides
- A Unified Contrastive Energy-based Model for Understanding the Generative Ability of Adversarial Training ICLR 2022 (π Silver Best Paper Award @ ICML 2021 AML workshop) 2022 PDF | Slides | Award
- Reparameterized Sampling for Generative Adversarial Networks ECML-PKDD 2021 2021 (π Best ML Paper Award (1/685). Invited to Machine Learning Journal) PDF | Code | Slides | Media | Talk | Award
- Equilibrium Image Denoising with Implicit Differentiation IEEE Trans. Image Processing (TIP) 2023 PDF
- A Message Passing Perspective on Learning Dynamics of Contrastive Learning ICLR 2023 2023 PDF
- Towards a Unified Theoretical Understanding of Non-contrastive Learning via Rank Differential Mechanism ICLR 2023 2023 PDF
- Rethinking the Effect of Data Augmentation in Adversarial Contrastive Learning ICLR 2023 2023 PDF | Code
- ContraNorm: A Contrastive Learning Perspective on Oversmoothing and Beyond ICLR 2023 2023 PDF
- Unbiased Stochastic Proximal Solver for Graph Neural Networks with Equilibrium States ICLR 2023 2023 PDF
- On the Connection between Invariant Learning and Adversarial Training for Out-of-Distribution Generalization AAAI 2023 (Oral) 2023 PDF
- How Mask Matters: Towards Theoretical Understandings of Masked Autoencoders NeurIPS 2022 (Spotlight, Top 5%) 2022 PDF | Code | Slides
- Improving Out-of-distribution Robustness by Adversarial Training with Structured Priors NeurIPS 2022 (Spotlight, Top 5%) 2022 PDF | Code | Slides
- When Adversarial Training Meets Vision Transformers: Recipes from Training to Architecture NeurIPS 2022 (Spotlight, Top 5%) 2022 PDF | Code
- Variational Energy-Based Models: A Probabilistic Framework for Contrastive Self-Supervised Learning NeurIPS 2022 SSL Workshop 2022 PDF
- AggNCE: Asymptotically Identifiable Contrastive Learning NeurIPS 2022 SSL Workshop (Oral) 2022 PDF
- Efficient and Scalable Implicit Graph Neural Networks with Virtual Equilibrium IEEE BigData 2022 (Long Talk) 2022 PDF
- Optimization-Induced Graph Implicit Nonlinear Diffusion ICML 2022 2022 PDF | Code
- G2CN: Graph Gaussian Convolution Networks with Concentrated Graph Filters ICML 2022 2022 PDF
- Chaos is a Ladder: A New Theoretical Understanding of Contrastive Learning via Augmentation Overlap ICLR 2022 2022 PDF | Code | Slides
- A Unified Contrastive Energy-based Model for Understanding the Generative Ability of Adversarial Training ICLR 2022 (π Silver Best Paper Award @ ICML 2021 AML workshop) 2022 PDF | Slides | Award
- Fooling Adversarial Training with Inducing Noise Tech report, Nov. 2021 2021 PDF
- Residual Relaxation for Multi-view Representation Learning NeurIPS 2021 2021 PDF | Slides | Blog
- Dissecting the Diffusion Process in Linear Graph Convolutional Networks NeurIPS 2021 2021 PDF | Code | Slides | Blog
- Reparameterized Sampling for Generative Adversarial Networks ECML-PKDD 2021 2021 (π Best ML Paper Award (1/685) & Invited to Machine Learning Journal) PDF | Code | Slides | Media | Talk | Award
- Train Once, and Decode as You Like COLING 2020 2020 PDF
- Decoder-free Robustness Disentanglement without (Additional) Supervision Tech report, July 2020 2020 PDF
- A Message Passing Perspective on Learning Dynamics of Contrastive Learning ICLR 2023 2023 PDF
- Towards a Unified Theoretical Understanding of Non-contrastive Learning via Rank Differential Mechanism ICLR 2023 2023 PDF
- Rethinking the Effect of Data Augmentation in Adversarial Contrastive Learning ICLR 2023 2023 PDF
- ContraNorm: A Contrastive Learning Perspective on Oversmoothing and Beyond ICLR 2023 2023 PDF
- How Mask Matters: Towards Theoretical Understandings of Masked Autoencoders NeurIPS 2022 (Spotlight, Top 5%) 2022 PDF | Code | Slides
- Variational Energy-Based Models: A Probabilistic Framework for Contrastive Self-Supervised Learning NeurIPS 2022 SSL Workshop 2022 PDF
- AggNCE: Asymptotically Identifiable Contrastive Learning NeurIPS 2022 SSL Workshop (Oral) 2022 PDF
- Chaos is a Ladder: A New Theoretical Understanding of Contrastive Learning via Augmentation Overlap ICLR 2022 2022 PDF | Code | Slides
- A Unified Contrastive Energy-based Model for Understanding the Generative Ability of Adversarial Training ICLR 2022 (π Silver Best Paper Award @ ICML 2021 AML workshop) 2022 PDF | Slides | Award
- Residual Relaxation for Multi-view Representation Learning NeurIPS 2021 2021 PDF | Slides | Blog
- Reparameterized Sampling for Generative Adversarial Networks ECML-PKDD 2021 2021 (π Best ML Paper Award (1/685). Invited to Machine Learning Journal) PDF | Code | Slides | Media | Talk | Award
- Rethinking the Effect of Data Augmentation in Adversarial Contrastive Learning ICLR 2023 2023 PDF
- On the Connection between Invariant Learning and Adversarial Training for Out-of-Distribution Generalization AAAI 2023 (Oral) 2023 PDF
- Improving Out-of-distribution Robustness by Adversarial Training with Structured Priors NeurIPS 2022 (Spotlight, Top 5%) 2022 PDF | Code | Slides
- When Adversarial Training Meets Vision Transformers: Recipes from Training to Architecture NeurIPS 2022 (Spotlight, Top 5%) 2022 PDF | Code
- A Unified Contrastive Energy-based Model for Understanding the Generative Ability of Adversarial Training ICLR 2022 (π Silver Best Paper Award @ ICML 2021 AML workshop) 2022 PDF | Slides | Award
- Fooling Adversarial Training with Inducing Noise Tech report, Nov. 2021 2021 PDF
- Decoder-free Robustness Disentanglement without (Additional) Supervision Tech report, July 2020 2020 PDF
- ContraNorm: A Contrastive Learning Perspective on Oversmoothing and Beyond ICLR 2023 2023 PDF
- Unbiased Stochastic Proximal Solver for Graph Neural Networks with Equilibrium States ICLR 2023 2023 PDF
- Efficient and Scalable Implicit Graph Neural Networks with Virtual Equilibrium IEEE BigData 2022 (Long Talk) 2022 PDF
- Optimization-Induced Graph Implicit Nonlinear Diffusion ICML 2022 2022 PDF | Code
- G2CN: Graph Gaussian Convolution Networks with Concentrated Graph Filters ICML 2022 2022 PDF
- Dissecting the Diffusion Process in Linear Graph Convolutional Networks NeurIPS 2021 2021 PDF | Code | Slides | Blog
Awards
Best ML Paper Award (1/685), ECML-PKDD, 2021
Silver Best Paper Award, ICML AML workshop, 2021
National Scholarship, Ministry of Education of China, 2021, 2022
Principal Scholarship, Peking University, 2022
Baidu Scholarship Nomination Award (20 worldwide), Baidu Inc, 2022
Silver Best Paper Award, ICML AML workshop, 2021
National Scholarship, Ministry of Education of China, 2021, 2022
Principal Scholarship, Peking University, 2022
Baidu Scholarship Nomination Award (20 worldwide), Baidu Inc, 2022
Professional Services
Reviewer for NeurIPS, ICLR, ICML, CVPR, ACL, EMNLP, ECML-PKDD