Yoshitomo Matsubara

Applied Scientist at Amazon | ML OSS developer
Ph.D. in Computer Science

Dr. Yoshitomo Matsubara is an Applied Scientist at Amazon and an ML OSS developer. He completed the Ph.D. program in Computer Science at University of California, Irvine (UCI) and worked on deep learning for resource-constrained edge computing systems with Profs. Marco Levorato, Stephan Mandt, and Sameer Singh. Before UCI, he obtained his Master and Bachelor degrees at University of Hyogo and National Institute of Technology, Akashi College, Japan, respectively.

His main research interests are in machine learning, natural language processing, computer vision, information retrieval, and symbolic regression. For deep learning, his main interests are in knowledge distillation and supervised compression. He is also a developer of ML OSS: torchdistill and sc2bench.
January 5, 2024
I am serving as the Technical Chair of Journal of Data-centric Machine Learning Research (DMLR).
January 4, 2024
My article about torchdistill was published in PyTorch's Medium. (Twitter/X, LinkedIn, Facebook)
December 5, 2023
My ML OSS, torchdistill, officially joined PyTorch Ecosystem.
November 20, 2023
I was selected as a Top Reviewer of NeurIPS 2023.
October 27, 2023
Our paper was accepted (Oral) at NeurIPS 2023 AI for Scientific Discovery: From Theory to Practice.
October 9, 2023
My paper was accepted at EMNLP 2023 Workshop for Natural Language Processing Open Source Software (NLP-OSS).
August 24, 2023
I am serving as the Technical Chair of CVPR 2024.
July 9, 2023
I am attending ACL 2023 at Toronto, Canada.
May 25, 2023
Our paper was accepted at Transactions on Machine Learning Research (TMLR).
May 2, 2023
Our paper was accepted at ACL 2023 as a long paper (Findings).
April 10, 2023
Our paper was accepted at ICDCS 2023.
March 7, 2023
I renovated my personal website.
November 29, 2022
I am attending NeurIPS 2022 at New Orleans, LA, USA.
November 1, 2022
I was selected as an Outstanding Reviewer of NeurIPS 2022 Datasets & Benchmarks track.
November 1, 2022
I was selected as a Top Reviewer of NeurIPS 2022 (main track).
October 20, 2022
Our paper was accepted at NeurIPS 2022 AI for Science: Progress and Promises.
October 6, 2022
Our paper was accepted at EMNLP 2022 as a long paper (Findings).
July 10, 2022
I am attending NAACL 2022 at Seattle, WA, USA.
April 21, 2022
I was selected as a Highlighted Reviewer of ICLR 2022.
March 28, 2022
I joined Amazon Alexa AI as an applied scientist.
March 23, 2022
I was selected as a Top Contributor of Papers with Code.
March 18, 2022
I completed my Ph.D. program at University of California, Irvine and obtained Ph.D. in Computer Science.
March 14, 2022
Our paper was accepted at ACM Computing Surveys (CSUR).
March 5, 2022
Our paper was accepted at IEEE WoWMoM 2022.
February 14, 2022
I successfully defended my Ph.D. thesis "Towards Split Computing: Supervised Compression for Resource-Constrained Edge Computing Systems".
January 13, 2022
I gave a tutorial "Reliable Real-Time Distributed AI for Mobile Autonomous Systems" with Prof. Marco Levorato at ICOIN 2022.
January 4, 2022
I am attending WACV 2022 at Waikoloa, HI, USA.
January 3, 2022
This quarter, I am working as a graduate student researcher.

simple distillation process, icon

Knowledge Distillation

Model compression and beyond.

arrow traffic sign with text OSS

Open Source Software

I develop and open-source (mostly ML) frameworks that boost my research and I hope would be useful for the research community too.

robots searching for data in digital world

Information Retrieval / Question Answering

Natural language processing for efficient IR and QA systems.

edge computing

Supervised Compression for Split Computing

Weak mobile device, limited communication capacity, strong edge server. How can we achieve efficient inference?

a person observes a big tree of apples, some apples are falling

Symbolic Regression for Scientific Discovery

Can we (re)discover hidden laws from observed data?

researchers carefully reading a research paper, illustration

Peer Review Systems

What are the problems in the current scientific peer review systems? How can we imporve the systems?