Hi👋 I'm Koki Maeda, currently 1st year of Ph.D course at Institute Science of Tokyo (Former Name: Tokyo Institute of Technology). I'm privileged to be advised by Prof Naoaki Okazaki and to have the opportunity to collaborate on research with Shuhei Kurita.

The aim of my research is to accurately evaluate the model’s understanding of the world. My work primarily focuses on developing robust evaluation metrics and methodologies to assess the performance of natural language processing and computer vision models in various real-world scenarios. I aim to contribute to the creation of more reliable and interpretable AI systems.

If you are interested in my work and have opportunities for research internships, visiting student positions, or collaborative research, please feel free to contact me:

E-mail -> koki.maeda[at]nlp[dot]c.titech.ac.jp

X -> @silviasetitech

• Vision and Language, especially scene understanding
• Evaluation
• (Vision) Language Models
• Grammatical Error Collection/Correction

1. Eri Onami, Taiki Miyanishi, Koki Maeda, and Shuhei Kurita. 2025. LegalViz: Legal Text Visualization by Text To Diagram Generation. 2025 Annual Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics (NAACL2025). Albuquerque, New Mexico. April 29–May 4. 2025.
2. Koki Maeda*, Tosho Hirasawa*, Atsushi Hashimoto, Jun Harashima, Leszek Rybicki, Yusuke Fukasawa, and Yoshitaka Ushiku. 2024. COM Kitchens: An Unedited Overhead-view Procedural Videos Dataset a Vision-Language Benchmark. In Proceedings of The 18th European Conference on Computer Vision: ECCV 2024, Milan, Italy. ECCV. (Acceptance Rate: 27.9%, *: Equally Contribution)
3. Koki Maeda, Shuhei Kurita, Taiki Miyanishi, and Naoaki Okazaki. 2023. Query-based Image Captioning from Multi-context 360$^/circ$ Images. In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 6940–6954, Singapore. Association for Computational Linguistics. (Acceptance Rate: 42.9%)
4. Taku Hasegawa, Kyosuke Nishida, Koki Maeda, and Kuniko Saito. 2023. DueT: Image-Text Contrastive Transfer Learning with Dual-adapter Tuning. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP2023), pages 13607–13624, Singapore. Association for Computational Linguistics. (Acceptance Rate: 21.3%)
5. Koki Maeda, Masahiro Kaneko, and Naoaki Okazaki. 2022. IMPARA: Impact-Based Metric for GEC Using Parallel Data. In Proceedings of the 29th International Conference on Computational Linguistics (COLING2022), pages 3578–3588, Gyeongju, Republic of Korea. International Committee on Computational Linguistics. (Acceptance Rate: 33.4%)

• 2024.04 - Present : Ph.D @ Institute Science of Tokyo (Former Name: Tokyo Institute of Technology), Tokyo, Japan
• 2022.04 - 2024.03 : Master @ Tokyo Institute of Technology, Tokyo, Japan
• 2018.04 - 2022.03 : Bachelor @ Tokyo Institute of Technology, Tokyo, Japan

• 2024.07 - Present Tokyo Institute of Technology, Research Assistant
  • Building Swallow, Japanese LLMs with our enthusiastic lab members!
• 2024.07 - Present Cierpa & Co., Engineering Intern
  • Working on research/developing document understanding system! 📔
• 2024.06 - Present NII, Research Assistant
  • Working on multimodal working group of LLM-jp 😏
• 2023.06 - 2024.01 OMRON SINIC X, Student Internship
• 2022.07 - 2024.03 RIKEN AIP, Research Part-time Worker II
• 2022.08 - 2022.09 NTT Lab., Summer Internship
• 2021.12 - 2022.12 Tokyo Institute of Technology, Research Assistant
• 2021.02 - 2022.03 Future, Member of Strategic AI Group

2024.04-2027.03 Tokyo Tech Program for Development of Co-creative Experts towards Top-level AI Research 3,600,000 JPY/year 2024.04-(canceled) Tokyo Tech Tsubame Scholarship for Doctoral Students 480,000 JPY / year