Gouki Minegishi

I’m a second-year PhD student at The University of Tokyo, mentored by Professor Yutaka Matsuo.

What I enjoy most is taking a model too complicated to reason about and finding a setting simple enough to explain completely — small models, controlled tasks, clean training regimes — then using what those settings reveal to work out how capabilities actually come about. The question underneath all of it is what human intelligence really is.

For more details, see my CV.

selected publications

ICML2026

Emergent Analogical Reasoning in Transformers

Gouki Minegishi, Jingyuan Feng, Hiroki Furuta, Takeshi Kojima, Yusuke Iwasawa, and 1 more author

In Forty-third International Conference on Machine Learning, ICML, 2026

Awarded arXiv Bib Code

Spotlight (top 2.2%)

@inproceedings{minegishi2026emergent,
  title = {Emergent Analogical Reasoning in Transformers},
  author = {Minegishi, Gouki and Feng, Jingyuan and Furuta, Hiroki and Kojima, Takeshi and Iwasawa, Yusuke and Matsuo, Yutaka},
  booktitle = {Forty-third International Conference on Machine Learning, ICML},
  year = {2026},
  url = {https://openreview.net/forum?id=aFCoTBGM4M},
  kind = {Conference},
}

ACL2026

Understanding Emergent Misalignment via Feature Superposition Geometry

Gouki Minegishi, Hiroki Furuta, Takeshi Kojima, Yusuke Iwasawa, and Yutaka Matsuo

In The 64th Annual Meeting of the Association for Computational Linguistics, 2026

arXiv Bib

@inproceedings{anonymous2026understanding,
  title = {Understanding Emergent Misalignment via Feature Superposition Geometry},
  author = {Minegishi, Gouki and Furuta, Hiroki and Kojima, Takeshi and Iwasawa, Yusuke and Matsuo, Yutaka},
  booktitle = {The 64th Annual Meeting of the Association for Computational Linguistics},
  year = {2026},
  url = {https://arxiv.org/abs/2605.00842},
  kind = {Conference},
}

Neurips2025

Topology of Reasoning: Understanding Large Reasoning Models through Reasoning Graph Properties

Gouki Minegishi, Hiroki Furuta, Takeshi Kojima, Yusuke Iwasawa, and Yutaka Matsuo

In The Thirty-ninth Annual Conference on Neural Information Processing Systems, Neurips, 2025

arXiv Bib

@inproceedings{minegishi2025topologyreasoningunderstandinglarge,
  title = {Topology of Reasoning: Understanding Large Reasoning Models through Reasoning Graph Properties},
  author = {Minegishi, Gouki and Furuta, Hiroki and Kojima, Takeshi and Iwasawa, Yusuke and Matsuo, Yutaka},
  booktitle = {The Thirty-ninth Annual Conference on Neural Information Processing Systems, Neurips},
  year = {2025},
  url = {https://openreview.net/forum?id=o1g8NWkxqf},
  kind = {Conference},
}

ICML2025

Beyond Induction Heads: In-Context Meta Learning Induces Multi-Phase Circuit Emergence

Gouki Minegishi, Hiroki Furuta, Shohei Taniguchi, Yusuke Iwasawa, and Yutaka Matsuo

In Forty-second International Conference on Machine Learning, ICML, 2025

arXiv Bib Code

@inproceedings{anonymous2025incontext,
  title = {Beyond Induction Heads: In-Context Meta Learning Induces Multi-Phase Circuit Emergence},
  author = {Minegishi, Gouki and Furuta, Hiroki and Taniguchi, Shohei and Iwasawa, Yusuke and Matsuo, Yutaka},
  booktitle = {Forty-second International Conference on Machine Learning, ICML},
  year = {2025},
  url = {https://openreview.net/forum?id=Xw01vF13aV},
  kind = {Conference},
}

ICLR2025

Rethinking Evaluation of Sparse Autoencoders through the Representation of Polysemous Words

Gouki Minegishi, Hiroki Furuta, Yusuke Iwasawa, and Yutaka Matsuo

In The Thirteenth International Conference on Learning Representations, ICLR, 2025

arXiv Bib Code

@inproceedings{minegishi2025rethinking,
  title = {Rethinking Evaluation of Sparse Autoencoders through the Representation of Polysemous Words},
  author = {Minegishi, Gouki and Furuta, Hiroki and Iwasawa, Yusuke and Matsuo, Yutaka},
  booktitle = {The Thirteenth International Conference on Learning Representations, ICLR},
  year = {2025},
  url = {https://openreview.net/forum?id=HpUs2EXjOl},
  kind = {Conference},
}