Publications

2026

1. Preprint

   Test-Time Optimization of Query Embeddings with Ranking Aware Reward Maximization

   Tianyu Chen, and Jiaxing Wu

   arXiv preprint arXiv:2608.12569. (Collaboration with Google DeepMind), 2026

   Distills ranking rewards into a learned vector in a frozen retriever’s embedding space: up to +8.36 nDCG@10 across 15 MTEB retrieval tasks, generalizing to unseen queries and tasks with no weight updates.

   Bib PDF

   @article{chen2026testtime,
     title = {Test-Time Optimization of Query Embeddings with Ranking Aware Reward Maximization},
     author = {Chen, Tianyu and Wu, Jiaxing},
     journal = {arXiv preprint arXiv:2608.12569. (Collaboration with Google DeepMind)},
     year = {2026},
     tldr = {Distills ranking rewards into a learned vector in a frozen retriever's embedding space: up to +8.36 nDCG@10 across 15 MTEB retrieval tasks, generalizing to unseen queries and tasks with no weight updates.}
   }

2. Preprint

   MT-EditFlow: Reinforcement Learning for Multi-Turn Image Editing with Flow Matching

   Jiahui Huang, Yasi Zhang, Tianyu Chen, and 6 more authors

   arXiv preprint arXiv:2606.01985, 2026

   Flow-matching RL for multi-turn image editing with multi-reward advantage broadcasting: +6.85 turn-3 overall score on FLUX.1-Kontext-dev, surpassing Qwen-Image-Edit.

   Bib PDF

   @article{huang2026mteditflow,
     title = {MT-EditFlow: Reinforcement Learning for Multi-Turn Image Editing with Flow Matching},
     author = {Huang, Jiahui and Zhang, Yasi and Chen, Tianyu and Wang, Shu and Xie, Jianwen and Leong, Oscar and Zhou, Mingyuan and Wang, Nanzhu and Wu, Ying Nian},
     journal = {arXiv preprint arXiv:2606.01985},
     year = {2026},
     tldr = {Flow-matching RL for multi-turn image editing with multi-reward advantage broadcasting: +6.85 turn-3 overall score on FLUX.1-Kontext-dev, surpassing Qwen-Image-Edit.}
   }

3. ICML 2026

   REAL: Regression-Aware Reinforcement Learning for LLM-as-a-Judge

   Tianyu Chen^*, Yasi Zhang^*, Mingyuan Zhou, and 3 more authors

   The 43rd International Conference on Machine Learning. (Collaboration with Google DeepMind), 2026

   Regression-aware policy gradients for LLM-as-a-Judge: +8.40 Pearson and +7.20 Spearman over the SFT baseline on Qwen3-32B, with stronger out-of-domain generalization.

   Bib PDF Code

   @article{chen2026real,
     title = {REAL: Regression-Aware Reinforcement Learning for LLM-as-a-Judge},
     author = {Chen, Tianyu and Zhang, Yasi and Zhou, Mingyuan and Leong, Oscar and Wu, Ying Nian and Lukasik, Michal},
     journal = {The 43rd International Conference on Machine Learning. (Collaboration with Google DeepMind)},
     year = {2026},
     tldr = {Regression-aware policy gradients for LLM-as-a-Judge: +8.40 Pearson and +7.20 Spearman over the SFT baseline on Qwen3-32B, with stronger out-of-domain generalization.}
   }

4. ICLR 2026

   EdiVal-Agent: An Object-Centric Framework for Automated, Scalable, Fine-Grained Evaluation of Multi-Turn Editing

   Tianyu Chen^*, Yasi Zhang^*, Zhi Zhang, and 13 more authors

   The 14th International Conference on Learning Representations, 2026

   An object-centric VLM agent that automates fine-grained, multi-turn image-editing evaluation without human annotators; EdiVal-Bench covers 9 instruction types and 16 editing models.

   Bib PDF Code Website

   @article{chen2026edival,
     title = {EdiVal-Agent: An Object-Centric Framework for Automated, Scalable, Fine-Grained Evaluation of Multi-Turn Editing},
     author = {Chen, Tianyu and Zhang, Yasi and Zhang, Zhi and Yu, Peiyu and Wang, Shu and Wang, Zhendong and Lin, Kevin and Wang, Xiaofei and Yang, Zhengyuan and Li, Linjie and Lin, Chung-Ching and Xie, Jianwen and Leong, Oscar and Wang, Lijuan and Wu, Ying Nian and Zhou, Mingyuan},
     journal = {The 14th International Conference on Learning Representations},
     year = {2026},
     media = {https://mp.weixin.qq.com/s/jlVDWl7wJ8SGFcGdRG7wGA},
     tldr = {An object-centric VLM agent that automates fine-grained, multi-turn image-editing evaluation without human annotators; EdiVal-Bench covers 9 instruction types and 16 editing models.}
   }

5. ICLR 2026

   Score Distillation Beyond Acceleration: Generative Modeling from Corrupted Data

   Tianyu Chen^*, Yasi Zhang^*, Zhendong Wang, and 3 more authors

   The 14th International Conference on Learning Representations, 2026

   Score distillation is not merely an acceleration technique — it enhances generation quality from corrupted data, both empirically and theoretically.

   Bib PDF

   @article{chen2026scoredistillation,
     title = {Score Distillation Beyond Acceleration: Generative Modeling from Corrupted Data},
     author = {Chen, Tianyu and Zhang, Yasi and Wang, Zhendong and Wu, Ying Nian and Zhou, Mingyuan and Leong, Oscar},
     journal = {The 14th International Conference on Learning Representations},
     year = {2026},
     tldr = {Score distillation is not merely an acceleration technique — it enhances generation quality from corrupted data, both empirically and theoretically.}
   }

6. ICML 2026

   Conformal C2ST: Turning Weak Classifiers into Strong Two-Sample Tests

   Tianyu Chen^*, Vansh Bansal^*, and James G. Scott

   The 43rd International Conference on Machine Learning, 2026

   Bib PDF

   @article{chen2026weakclassifiers,
     title = {Conformal C2ST: Turning Weak Classifiers into Strong Two-Sample Tests},
     author = {Chen, Tianyu and Bansal, Vansh and Scott, James G.},
     journal = {The 43rd International Conference on Machine Learning},
     year = {2026},
   }

2025

1. Preprint

   Restoration score distillation: From corrupted diffusion pretraining to one-step high-quality generation

   Yasi Zhang^*, Tianyu Chen^*, Zhendong Wang, and 3 more authors

   arXiv preprint arXiv:2505.13377, 2025

   Bib PDF

   @article{zhang2025restoration,
     title = {Restoration score distillation: From corrupted diffusion pretraining to one-step high-quality generation},
     author = {Zhang, Yasi and Chen, Tianyu and Wang, Zhendong and Wu, Ying Nian and Zhou, Mingyuan and Leong, Oscar},
     journal = {arXiv preprint arXiv:2505.13377},
     year = {2025},
   }

2. Preprint

   A Generative Framework for Causal Estimation via Importance-Weighted Diffusion Distillation

   Tianyu Chen^*, Xinran Song^*, and Mingyuan Zhou

   Preprint., 2025

   Bib PDF

   @article{chen2026causal,
     title = {A Generative Framework for Causal Estimation via Importance-Weighted Diffusion Distillation},
     author = {Chen, Tianyu and Song, Xinran and Zhou, Mingyuan},
     journal = {Preprint.},
     year = {2025},
   }

3. NeurIPS 2025

   Improving Data Efficiency for LLM Reinforcement Fine-tuning Through Difficulty-targeted Online Data Selection and Rollout Replay

   Yifan Sun, Jingyan Shen, Yibin Wang, and 4 more authors

   Advances in Neural Information Processing Systems 2025, 2025

   Difficulty-targeted online data selection and rollout replay cut GRPO fine-tuning time by 23-62% while matching the original performance.

   Bib PDF

   @article{sun2025improving,
     title = {Improving Data Efficiency for LLM Reinforcement Fine-tuning Through Difficulty-targeted Online Data Selection and Rollout Replay},
     author = {Sun, Yifan and Shen, Jingyan and Wang, Yibin and Chen, Tianyu and Wang, Zhendong and Zhou, Mingyuan and Zhang, Huan},
     journal = {Advances in Neural Information Processing Systems 2025},
     year = {2025},
     tldr = {Difficulty-targeted online data selection and rollout replay cut GRPO fine-tuning time by 23-62% while matching the original performance.}
   }

4. NeurIPS 2025

   CoLT: The conditional localization test for assessing the accuracy of neural posterior estimates

   Tianyu Chen, Vansh Bansal, and James G. Scott

   Advances in Neural Information Processing Systems 2025, 2025

   A principled conditional localization test for validating neural posterior estimators. NeurIPS 2025 Spotlight.

   Spotlight Bib PDF Code

   Selected as a Spotlight paper at NeurIPS 2025.

   @article{chen2025colt,
     title = {CoLT: The conditional localization test for assessing the accuracy of neural posterior estimates},
     author = {Chen, Tianyu and Bansal, Vansh and Scott, James G.},
     journal = {Advances in Neural Information Processing Systems 2025},
     year = {2025},
     tldr = {A principled conditional localization test for validating neural posterior estimators. NeurIPS 2025 Spotlight.}
   }

5. AISTATS 2025

   Conditional diffusions for amortized neural posterior estimation

   Tianyu Chen, Vansh Bansal, and James G. Scott

   The 28th International Conference on Artificial Intelligence and Statistics., 2025

   Diffusion is a good tool to do posterior sampling.

   Bib PDF Code

   @article{chen2024conditional,
     title = {Conditional diffusions for amortized neural posterior estimation},
     author = {Chen, Tianyu and Bansal, Vansh and Scott, James G.},
     journal = {The 28th International Conference on Artificial Intelligence and Statistics.},
     year = {2025},
     tldr = {Diffusion is a good tool to do posterior sampling.},
   }

2024

1. Preprint

   Enhancing and Accelerating Diffusion-Based Inverse Problem Solving through Measurements Optimization

   Tianyu Chen, Zhendong Wang, and Mingyuan. Zhou

   Submitted., 2024
2. NeurIPS 2024

   Diffusion Policies creating a Trust Region for Offline Reinforcement Learning

   Tianyu Chen, Zhendong Wang, and Mingyuan Zhou

   Advances in Neural Information Processing Systems 2024, 2024

   Diffusion loss can be used to distill one-step policy and encourage mode-seeking.

   Bib PDF Code

   @article{chen2024diffusion,
     title = {Diffusion Policies creating a Trust Region for Offline Reinforcement Learning},
     author = {Chen, Tianyu and Wang, Zhendong and Zhou, Mingyuan},
     journal = {Advances in Neural Information Processing Systems 2024},
     year = {2024},
     url = {https://arxiv.org/abs/2405.19690},
     tldr = {Diffusion loss can be used to distill one-step policy and encourage mode-seeking.}
   }

3. NeurIPS 2024

   Identifying General Mechanism Shifts in Linear Causal Representations

   Tianyu Chen, Kevin Bello, Francesco Locatello, and 2 more authors

   Advances in Neural Information Processing Systems 2024, 2024

   Bib PDF Code

   @article{chen2024identifying,
     title = {Identifying General Mechanism Shifts in Linear Causal Representations},
     author = {Chen, Tianyu and Bello, Kevin and Locatello, Francesco and Aragam, Bryon and Ravikumar, Pradeep},
     journal = {Advances in Neural Information Processing Systems 2024},
     year = {2024},
   }

4. PNAS

   Joint trajectory inference for single-cell genomics using deep learning with a mixture prior

   Tianyu Chen^*, Jin-Hong Du^*, Ming Gao, and 1 more author

   Proceedings of the National Academy of Sciences, 2024

   A VAE with a hierarchical prior offers a comprehensive pipeline for integrating multi-omic data, correcting batch effects, inferring pseudotime, and conducting differential analysis.

   Bib PDF Code

   @article{du2020model,
     title = {Joint trajectory inference for single-cell genomics using deep learning with a mixture prior},
     author = {Chen, Tianyu and Du, Jin-Hong and Gao, Ming and Wang, Jingshu},
     journal = {Proceedings of the National Academy of Sciences},
     pages = {Vol. 121 | No. 37},
     year = {2024},
     url = {https://www.pnas.org/doi/10.1073/pnas.2316256121},
     tldr = {A VAE with a hierarchical prior offers a comprehensive pipeline for integrating multi-omic data, correcting batch effects, inferring pseudotime, and conducting differential analysis.}
   }

2023

1. NeurIPS 2023

   iSCAN: identifying causal mechanism shifts among nonlinear additive noise models

   Tianyu Chen, Kevin Bello, Bryon Aragam, and 1 more author

   Advances in Neural Information Processing Systems 2023, 2023

   Score matching help directly detect shifted nodes among different graphs.

   Bib PDF Code

   @article{chen2024iscan,
     title = {iSCAN: identifying causal mechanism shifts among nonlinear additive noise models},
     author = {Chen, Tianyu and Bello, Kevin and Aragam, Bryon and Ravikumar, Pradeep},
     journal = {Advances in Neural Information Processing Systems 2023},
     volume = {36},
     year = {2023},
     url = {https://arxiv.org/abs/2306.17361},
     tldr = {Score matching help directly detect shifted nodes among different graphs.}
   }

2022

1. Book Chapter

   Deep Learning Methods for Single-Cell Omics Data

   Jingshu Wang, and Tianyu Chen

   In Handbook of Statistical Bioinformatics, 2022

   An overview of machine learning methods in single cell data.

   Bib PDF

   @incollection{wang2022deep,
     title = {Deep Learning Methods for Single-Cell Omics Data},
     author = {Wang, Jingshu and Chen, Tianyu},
     booktitle = {Handbook of Statistical Bioinformatics},
     pages = {109--132},
     year = {2022},
     publisher = {Springer},
     tldr = {An overview of machine learning methods in single cell data.},
     url = {https://link.springer.com/chapter/10.1007/978-3-662-65902-1_6},
   }