About

Postdoctoral Researcher at MBZUAI, Visiting Scientist at RIKEN AIP.
My research focuses on understanding how LLMs acquire knowledge from training data, interpret it internally, and leverage it for reasoning.
My CV is here (last updated: Oct. 16, 2024).

Work Experience

2025/05 –
NLU Team, RIKEN AIP
Visiting Scientist
2024/10 –
Kentaro Team, MBZUAI
Researcher
2023/04 – 2024/09
Megagon Labs
Research Scientist
2023/06 – 2024/09
2021/04 – 2023/03
JSPS DC2
Research Fellow
2019/06 – 2023/03
Megagon Labs
Research Intern
2018/10 – 2023/03
CyberAgent, Inc.
Collaborative Project
2018/10 – 2023/03

Publications

Journal (Refereed)

  • Nearest Neighbor Non-autoregressive Text Generation Journal of Information Processing, 2023 J-STAGE
    Ayana Niwa, Sho Takase, Naoaki Okazaki
  • Construction of a Corpus of Rhetorical Devices in Slogans and Structural Analysis of Antitheses ACM TALLIP, 2021 ACM
    Ayana Niwa, Naoaki Okazaki, Kohei Wakimoto, Keisuke Nishiguchi, Masataka Mouri

International Conference (Refereed)

  • Rectifying Belief Space via Unlearning to Harness LLMs' Reasoning ACL 2025 Findings Anthology
    Ayana Niwa, Masahiro Kaneko, Kentaro Inui
  • AmbigNLG: Addressing Task Ambiguity in Instruction for NLG EMNLP 2024Anthology
    Ayana Niwa, Hayate Iso
  • Interpretability for Language Learners Using Example-Based GEC ACL 2022 Anthology
    Masahiro Kaneko, Sho Takase, Ayana Niwa, Naoaki Okazaki
  • Predicting Antonyms in Context using BERT INLG 2021 Anthology
    Ayana Niwa, Keisuke Nishiguchi, Naoaki Okazaki

Article

  • NLP Colloquium Journal of Natural Language Processing, 2024 J-STAGE
    Ayana Niwa, Sho Yokoi, Junya Takayama, Itsumi Saito

Domestic Conference and Symposium (Unrefereed)

  • Addressing Ambiguity in Instruction for NLG NLP 2024, Hyogo, Japan PDF
    Ayana Niwa, Hayate Iso

Education

2020/04 – 2023/03
Ph.D., Tokyo Institute of TechnologyAdviser: Prof. Naoaki Okazaki
2018/04 – 2020/03
M.Eng., Tokyo Institute of TechnologyAdviser: Prof. Naoaki Okazaki
2014/04 – 2018/03
B.S., Sophia UniversityAdviser: Prof. Takashi Irohara

Talks

  • “A Quick Overview to Unlock the Potential of LLMs through Prompt Engineering.” Tokyo AI, 2024/08 Slides
  • “The Frontiers of Mathematics, Information, and AI.” RIKEN AIP, 2022/11 YouTube
  • “Career Path as a Female Researcher.” NeurIPS Meetup Japan WiML, 2021/12
  • “Derivation and Use of Templates in NLG.”SIG‑FPAI, 2021/11 Info
  • “Trends in NLP at NeurIPS 2019.”JSAI AI Seminar, 2020/03 Slides
  • “A Smart Computer that Understands Language.”A visiting lecture at Jissen Gakuen Senior High School, 2019/11
  • “Natural Language Processing.”A visiting lecture at Denenchofu High School, 2019/06

Committee

Others

Scholarships & Grants

  • JSPS Research Fellowship (DC2) 4.8M JPY/yr (2021–2023) [Acceptance Rate < 20%]
  • JSPS Fellows Grant (21J13602) 1.5M JPY (2021–2023)
  • NeurIPS 2019 travel & registration supportJSAI
  • Tokyo Tech Tsubame Scholarship480k JPY/yr (2020–2021)

Awards

  • Young Researcher’s Encouragement Award NLP 2024
  • Committee Special Award NLP 2022
  • Nomination Award 2021 Microsoft Research Asia Fellowship

Teaching & Mentoring

  • Mentor, UGRIP program, MBZUAI (2025)two mentee projects; one Best Presentation Award🎉
  • TA, Machine Learning, Tokyo Tech 2020 4Q, 2021 4Q