2026

1. Thinking by Subtraction: Confidence-Driven Contrastive Decoding for LLM Reasoning

   Lexiang Tang, Weihao Gao, Bingchen Zhao, and 3 more authors

   Feb 2026

   arXiv Bib

   @misc{tang2026thinking,
     title = {Thinking by Subtraction: Confidence-Driven Contrastive Decoding for LLM Reasoning},
     author = {Tang, Lexiang and Gao, Weihao and Zhao, Bingchen and Ma, Lu and Yang, Bang and Zou, Yuexian},
     year = {2026},
     month = feb,
     journal = {arXiv preprint arXiv:2602.18232},
   }

2. GIFT: Unlocking Global Optimality in Post-Training via Finite-Temperature Gibbs Initialization

   Zhengyang Zhao†, Lu Ma†, Yizhen Jiang, and 7 more authors

   Jan 2026

   arXiv Bib

   @misc{zhao2026gift,
     title = {{GIFT}: Unlocking Global Optimality in Post-Training via Finite-Temperature Gibbs Initialization},
     author = {Zhao†, Zhengyang and Ma†, Lu and Jiang, Yizhen and Ma, Xiaochen and Meng, Zimo and Shen, Chengyu and Tang, Lexiang and Sun, Haoze and Pei, Peng and Zhang, Wentao},
     year = {2026},
     month = jan,
     journal = {arXiv preprint arXiv:2601.09233},
   }

3. Leash: Adaptive Length Penalty and Reward Shaping for Efficient Large Reasoning Model

   Yanhao Li†, Lu Ma†, Jiaran Zhang, and 3 more authors

   ACL 2026

   arXiv Bib

   @inproceedings{li2025leash,
     title = {Leash: Adaptive Length Penalty and Reward Shaping for Efficient Large Reasoning Model},
     author = {Li†, Yanhao and Ma†, Lu and Zhang, Jiaran and Tang, Lexiang and Zhang, Wentao and Luo, Guibo},
     year = {2026},
     month = dec,
     booktitle = {ACL 2026},
   }

2025

1. DataFlow: An LLM-Driven Framework for Unified Data Preparation and Workflow Automation in the Era of Data-Centric AI

   Hao Liang, Xiaochen Ma, Zhou Liu, and 31 more authors

   Dec 2025

   arXiv Bib

   @misc{liang2025dataflow,
     title = {{DataFlow}: An {LLM}-Driven Framework for Unified Data Preparation and Workflow Automation in the Era of Data-Centric {AI}},
     author = {Liang, Hao and Ma, Xiaochen and Liu, Zhou and Wong, Zhen Hao and Zhao, Zhengyang and Meng, Zimo and He, Runming and Shen, Chengyu and Cai, Qifeng and Han, Zhaoyang and Qiang, Meiyi and Feng, Yalin and Bai, Tianyi and Pan, Zewei and Guo, Ziyi and Jiang, Yizhen and Deng, Jingwen and You, Qijie and Lai, Peichao and Guo, Tianyu and Tsai, Chi Hsu and Feng, Hengyi and Hu, Rui and Yu, Wenkai and Niu, Junbo and Zeng, Bohan and An, Ruichuan and Ma, Lu and Huang, Jihao and Zheng, Yaowei and He, Conghui and Tang, Linpeng and Cui, Bin and Zhang, Wentao},
     year = {2025},
     month = dec,
     journal = {arXiv preprint arXiv:2512.16676},
   }

2. DARO: Difficulty-Aware Reweighting Policy Optimization

   Jingyu Zhou†, Lu Ma†, Hao Liang, and 3 more authors

   Oct 2025

   arXiv Bib

   @misc{zhou2025daro,
     title = {{DARO}: Difficulty-Aware Reweighting Policy Optimization},
     author = {Zhou†, Jingyu and Ma†, Lu and Liang, Hao and Shen, Chengyu and Cui, Bin and Zhang, Wentao},
     year = {2025},
     month = oct,
     journal = {arXiv preprint arXiv:2510.09001},
   }

3. Native Visual Understanding: Resolving Resolution Dilemmas in Vision-Language Models

   Junbo Niu, Yuanhong Zheng, Ziyang Miao, and 8 more authors

   CVPR 2025

   arXiv Bib

   @inproceedings{niu2025native,
     title = {Native Visual Understanding: Resolving Resolution Dilemmas in Vision-Language Models},
     author = {Niu, Junbo and Zheng, Yuanhong and Miao, Ziyang and Dong, Hejun and Ge, Chunjiang and Liang, Hao and Ma, Lu and Zeng, Bohan and Zheng, Qiahao and He, Conghui and Zhang, Wentao},
     year = {2025},
     month = jun,
     booktitle = {CVPR 2025},
   }

4. Not All Tokens and Heads Are Equally Important: Dual-Level Attention Intervention for Hallucination Mitigation

   Lexiang Tang, Xianwei Zhuang, Bang Yang, and 5 more authors

   AAAI 2025

   Bib

   @inproceedings{tang2025notall,
     title = {Not All Tokens and Heads Are Equally Important: Dual-Level Attention Intervention for Hallucination Mitigation},
     author = {Tang, Lexiang and Zhuang, Xianwei and Yang, Bang and Hu, Zhiyuan and Li, Hongxiang and Ma, Lu and Ru, Jinghan and Zou, Yuexian},
     year = {2025},
     month = jun,
     booktitle = {AAAI 2025},
   }

5. Learning What Reinforcement Learning Can’t: Interleaved Online Fine-Tuning for Hardest Questions

   Lu Ma, Hao Liang, Meiyi Qiang, and 9 more authors

   ICLR 2026

   Bib

   @inproceedings{ma2025learning,
     title = {Learning What Reinforcement Learning Can't: Interleaved Online Fine-Tuning for Hardest Questions},
     author = {Ma, Lu and Liang, Hao and Qiang, Meiyi and Tang, Lexiang and Ma, Xiaochen and Wong, Zhen Hao and Niu, Junbo and Shen, Chengyu and He, Runming and Li, Yanhao and Cui, Bin and Zhang, Wentao},
     year = {2025},
     month = jun,
     booktitle = {ICLR 2026},
   }

6. Acceleration Algorithms in GNNs: A Survey

   Lu Ma, Zeang Sheng, Xunkai Li, and 7 more authors

   IEEE Transactions on Knowledge and Data Engineering

   Bib

   @article{ma2025acceleration,
     title = {Acceleration Algorithms in {GNN}s: A Survey},
     author = {Ma, Lu and Sheng, Zeang and Li, Xunkai and Gao, Xinyi and Hao, Zhezheng and Yang, Ling and Nie, Xiaonan and Jiang, Jiawei and Zhang, Wentao and Cui, Bin},
     year = {2025},
     month = feb,
     journal = {IEEE Transactions on Knowledge and Data Engineering},
     publisher = {IEEE},
   }