Scholar
Masato Hagiwara
Google Scholar ID: DepvCfQAAAAJ
Earth Species Project
Citations & Impact
Citations
751
H-index
13
i10-index
19
Publications
20
Co-authors
8
list available
Publications
Resume (English only)
Academic Achievements
- Paper on NatureLM-audio accepted at ICLR 2025—the first large audio-language model for animal sounds.
- Project MOSLA paper accepted at LREC-COLING 2024 and nominated for Best Paper.
- ISPA paper accepted at XAI-SA Workshop (ICASSP 2024).
- Co-authored AVES and BEANS papers accepted at ICASSP 2023.
- GitHub Typo Corpus paper accepted at LREC 2020.
- TEASPN framework paper accepted at EMNLP 2019 (system demonstration).
- Research featured in TechCrunch and Quartz.
- Developed GrammarTagger (neural multilingual grammar profiler) and EXPATS (explainable automated text scoring toolkit).
Research Experience
- Research Lead at Earth Species Project since November 2021, leading projects on non-human communication decoding.
- Former Machine Learning Engineer/Researcher at Duolingo, led the launch of Japanese, Korean, and Chinese courses.
- Collaborated with Mirai Translate (a Japan-based MT startup) and ACTNext (ACT’s R&D unit for educational research).
- Co-developed an ultra fine-grained NER system with Studio Ousia, ranked #2 at TAC KBP 2019.
- Co-authored the book 'Real-World Natural Language Processing' and is writing a book on Japanese NLP with Paul O'Leary McCann.
Background
- Currently a Research Lead at Earth Species Project, working on decoding non-human communication using AI/ML technologies.
- Pioneering the field of Animal Language Processing (ALP), focusing on foundation models (e.g., NatureLM-audio, AVES) and benchmarks (e.g., BEANS) for non-human animals.
- Fluent in Chinese, Japanese, and English; currently learning Korean and Lojban.
- Passionate about connecting language and machine learning to help people learn languages.
- Diagnosed with stage IV lung cancer in March 2023; currently stable but with a poor prognosis; raising funds for family and medical expenses.
