Yige Yuan

(袁一歌)

Introduction

I am currently pursuing my Ph.D. at the CAS Key Laboratory of AI Safety, advised by Prof. Xueqi Cheng and Assoc. Prof. Bingbing Xu. I obtained my B.S. degree in Information Security from Xidian University in 2020.

My research goal is to build Trustworthy AI that performs reliably across diverse scenarios.

To achieve this, I worked on Generalization & Robustness and Alignment & Hallucination within domains of graph, vision, and language, for tasks of both discriminative and generative modeling.

Experience

• Ph.D., Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China. 2020 - Now
  Advisor: Prof. Xueqi Cheng and Assoc. Prof. Bingbing Xu
  Ph.D. Topic: Machine Learning Generalization
• B.S., Department of Network and Information Security, Xidian University, Xian, China. 2016 - 2020
  Bachelor of Information Security (Experimental Class)

Publications

1. TEA: Test-time Energy Adaptation
   
   Yige Yuan, Bingbing Xu, Liang Hou, Fei Sun, Huawei Shen, Xueqi Cheng
   

   C IEEE/CVF Computer Vision and Pattern Recognition Conference (CVPR 2024)

   Seattle WA, USA; Main Track; CCF-A

   [Paper] [Code] [Slides] [Summary]

   

   TL;DR: In this paper, we propose to investigate generalization from an energy-based perspective and introduce TEA, a test-time adaptation method which transforms the trained classifier into an energy-based model and aligns the model's distribution with the test data's, enhancing its ability to perceive test distributions and thus improving overall generalizability.

2. PDE+: Enhancing Generalization via PDE with Adaptive Distributional Diffusion
   
   Yige Yuan, Bingbing Xu, Bo Lin, Liang Hou, Fei Sun, Huawei Shen, Xueqi Cheng
   

   C Annual AAAI Conference on Artificial Intelligence (AAAI 2024)

   Vancouver, Canada; Main Track; CCF-A

   [Paper] [Code] [Poster] [Slides] [Video] [Summary]

   

   TL;DR: In this paper, we propose to investigate generalization from PDE perspective and propose PDE-ADD framework. We introduce adaptive distributional diffusion into transport equation to enhance smoothness of its solution, thereby improving generalization directly via the underlying function of NN.

3. Towards Generalizable Graph Contrastive Learning: An Information Theory Perspective
   
   Yige Yuan, Bingbing Xu, Huawei Shen, Qi Cao, Keting Cen, Wen Zheng, Xueqi Cheng
   

   J Neural Networks (NN), Volume 172, April 2024

   Elsevier; IF=8.4, Q1, CCF-B

   [Paper] [Code] [Summary]

   

   TL;DR: In this paper, we propose a GCL generalization ability metric and prove a MI upper bound for it from an information-theoretic perspective. Guided by the bound, we design an InfoAdv framework, which can be applied to current GCL models and achieves SOTA performance.

4. Cal-DPO: Calibrated Direct Preference Optimization for Language Model Alignment
   
   Teng Xiao, Yige Yuan, Huaisheng Zhu, Mingxiao Li, Vasant G Honavar
   

   C Annual Conference on Neural Information Processing Systems (NeurIPS 2024)

   Vancouver, Canada; Main Track; CCF-A

   [Summary]

   

   TL;DR: In this paper, we introduce

5. How to Leverage Demonstration Data in Alignment for Large Language Model? A Self-Imitation Learning Perspective
   
   Teng Xiao, Mingxiao Li, Yige Yuan, Huaisheng Zhu, Chao Cui, Vasant G Honavar
   

   C The 2024 Conference on Empirical Methods in Natural Language Processing (EMNLP 2024)

   Miami, Florida, USA; Main Paper; CCF-B

   [Summary]

   

   TL;DR: In this paper, we introduce the Generalized Self-Imitation Learning (GSIL) framework. GSIL derives a surrogate objective for imitation learning through density ratio estimation. This surrogate objective: (1) eliminates the need for complex adversarial training in imitation learning, facilitating lightweight fine-tuning for LLMs; (2) encompasses a family of offline losses, offering a unified approach to alignment with demonstration data.

6. Negative as Positive: Enhancing Out-of-distribution Generalization for Graph Contrastive Learning
   
   Zixu Wang, Bingbing Xu, Yige Yuan, Huawei Shen and Xueqi Cheng
   

   C International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR 2024)

   Washington D.C., USA; Short Paper; CCF-A

   [Paper] [Summary]

   

   TL;DR: In this paper, we point out the traditional InfoNCE inevitably enlarges the distribution gap between different domains and impairs OOD generalization, due to it restricts the cross-domain pairs only to be negative samples. We treat the most semantically similar cross-domain negative pairs as positive to address this issue.

7. Augmentation-Aware Self-Supervision for Data-Efficient GAN Training
   
   Liang Hou, Qi Cao, Yige Yuan, Songtao Zhao, Chongyang Ma, Siyuan Pan, Pengfei Wan, Zhongyuan Wang, Huawei Shen, Xueqi Cheng
   

   C Annual Conference on Neural Information Processing Systems (NeurIPS 2023)

   New Orleans, USA; Main Track; CCF-A

   [Paper] [Code] [Poster] [Slides] [Video] [Summary]

   

   TL;DR: This paper proposes an augmentation-aware self-supervised discriminator to improve the data efficiency of training GANs.

8. History Driven Sampling for Scalable Graph Neural Networks
   
   Yang Li, Bingbing Xu, Fei Sun, Qi Cao, Yige Yuan, and Huawei Shen
   

   C International Conference on Database Systems for Advanced Applications (DASFAA 2024)

   Gifu, Japan; Reseach Track; CCF-B

   [Paper] [Summary]

   

   TL;DR: In this paper, we identify circular dependency issue of GNNs, i.e., the sampling probability of the minimum variance sampler is determined by node embeddings, while node embeddings can not be calculated until sampling is finished. We propose HDSGNN, which estimates the minimum variance sampler with the historical nodes' embeddings to break the circular dependency for better scalability of GNNs.

Awards && Honors

• Presidential Scholarship, Institute of Computing Technology 2023
• First-Class Scholarship, University of Chinese Academy of Sciences 2022
• Outstanding Student Award, University of Chinese Academy of Sciences 2022
• First Prize, 12th National College Students Information Security Contest 2019
• First Prize, 15th National Science and Technology Academic Competition of Challenge Cup 2017

Invited Talks

• LOGS Webinar, 2024.03 [video]

Academic Services

• Conference Reviewers: NeurIPS 2024, EMNLP 2024, ICLR 2025, AISTATS 2025, MIDL 2025, ACL Rolling Review
• Journal Reviewers: IEEE Transactions on Knowledge and Data Engineering (TKDE)