Thanawat Lodkaew (Sky)

ロードケオ　タナワット / ธนวัฒน์ หลอดแก้ว (สกาย)
Contact: lodkaew [at] ms.k.u-tokyo [dot] ac [dot] jp

About

Hello! I’m Thanawat or Sky. I’m a PhD student at The University of Tokyo under the supervision of Prof. Takashi Ishida and Prof. Masashi Sugiyama.

I’m interested in reliable LLM evaluation, robust learning, and understanding reward hacking in language models.

Publications

Conference Papers

1. How Can I Publish My LLM Benchmark Without Giving the True Answers Away? Takashi Ishida, Thanawat Lodkaew, and Ikko Yamane International Conference on Machine Learning (ICML), 2026, to appear.
[paper] [code] [dataset] [blog] Oral presentation at MemFM@ICML 2025 workshop.
2. Hybrid Loss for Improving Classification Performance with Unbalanced Data Thanawat Lodkaew, and Kitsuchart Pasupa International Conference on Neural Information Processing (ICONIP), 2020.
[paper] [code]
3. A New Approach to Automatic Heat Detection of Cattle in Video Kitsuchart Pasupa, and Thanawat Lodkaew International Conference on Neural Information Processing (ICONIP), 2019.
[paper]
4. Fashion Finder: A System for Locating Online Stores on Instagram from Product Images Thanawat Lodkaew, Weeruhputt Supsohmboon, Kitsuchart Pasupa, and Chu Kiong Loo International Conference on Information Technology and Electrical Engineering (ICITEE), 2018.
[paper] [video]

Journal Papers

1. On Symmetric Losses for Robust Policy Optimization with Noisy Preferences Soichiro Nishimori, Yu-Jie Zhang, Thanawat Lodkaew, and Masashi Sugiyama Transactions on Machine Learning Research (TMLR), 2026.
[paper] [code]
2. Importance Weighting for Aligning Language Models under Deployment Distribution Shift Thanawat Lodkaew, Tongtong Fang, Takashi Ishida, and Masashi Sugiyama Transactions on Machine Learning Research (TMLR), 2025.
[paper] [code] [poster] Achieved expert certification!
3. CowXNet: An Automated Cow Estrus Detection System Thanawat Lodkaew, Kitsuchart Pasupa, and Chu Kiong Loo Expert Systems with Applications, 2022.
[paper] [video] [code]
4. Visual Goal Human-Robot Communication Framework with Few-Shot Learning: A Case Study in Robot Waiter System Guntitat Sawadwuthikul, Tanyatep Tothong, Thanawat Lodkaew, Puchong Soisudarat, Sarana Nutanong, Poramate Manoonpong, and Nat Dilokthanakul IEEE Transactions on Industrial Informatics, 2021.
[paper]
This page is modified from Avicenna Theme | Built with Hugo