Electric Power to the People
Energy/Data Scientist at Oxford | Toyota
About Me
Masaki Adachi (足立 真輝)🇯🇵🇬🇧
I am a final-year DPhil student in Engineering Science (Information Engineering) in the Bayesian Exploration Lab | Machine Learning Reading Group at the University of Oxford under primary supervision of Prof Michael Osborne. My secondary advisor is Prof David Howey from Battery Intelligence Lab. I am also a Clarendon Scholar. Simultaneously, I am an assistant manager and data scientist at Toyota Motor Corporation. My research interest is in the interdisciplinary field between machine learning and energy science, applying Bayesian machine learning to the proliferation of electric vehicles and renewable energy.
Research Interests
-
Gaussian Process, Bayesian Quadrature, Bayesian Optimization, Probabilistic Numerics
-
Simulation-based inference, Battery modeling (physics-based, data-driven)
Recent News
-
[10.2024]
-
A new preprint is now on arXiv.
-
"Demonstrating Linked Battery Data To Accelerate Knowledge Flow in Battery Science" [arXiv]
-
-
Our work was presented at the Fractionation of isotopes in space II: from the Solar System to galaxies
-
"A Bayesian Approach for Fast Visualisation and Analysis of Broad-band Line-rich Datacubes"
-
-
I gave an invited talk at the following institution:
-
Automatic Control Laboratory, EPFL, hosted by Prof. Colin N Jones.
-
-
-
[09.2024]
Two papers are accepted in NeurIPS 2024.
-
[07.2024]
-
I gave a poster presentation at ELLIS Robust LLMs Workshop.
-
I gave two invited talks at the following institutions:
-
Max Planck Institute for extraterrestrial physics, hosted by Dr. Lin Yuxin.
-
Imperial College London, TLDR group, hosted by Prof. Sam Cooper.
-
-
-
[06.2024]
I was invited and visited the Max Planck Institute for extraterrestrial physics, the lab led by Dr. Silvia Spezzano. -
[05.2024]
Two papers are presented in AISTATS 2024, Valencia, Spain. -
[04.2024]
-
Started the internship at CISPA with Dr. Krikamol Muandet as an adviser.
-
A new preprint "A Quadrature Approach for General-Purpose Batch Bayesian Optimization via Probabilistic Lifting" is now on arXiv [arXiv] [code]
-
I gave a poster presentation at Oxford Battery Modelling Symposium.
-
I gave two invited talks at the following institutions:
-
RIKEN AIP, Japan, hosted by Prof. Emtiyaz Khan. [link]
-
CISPA, Germany, hosted by Dr. Krikamol Muandet
-
-
-
[02.2024]
-
A new preprint "Beyond Lengthscales: No-regret Bayesian Optimisation With Unknown Hyperparameters Of Any Type" is now on arXiv. [arXiv]
-
I gave an invited talk for LAM Research, CA, USA, hosted by Dr. Joe Lu.
-
-
[07.2023]
-
SOBER was accepted at ICML SODS workshop [openreview] [arXiv] [GitHub]
-
I gave an oral presentation at IFAC2023 Yokohama
-
Selection Bayesian Model Selection of Lithium-Ion Battery Models via Bayesian Quadrature [Journal (Open access)][arXiv] [GitHub]
-
-
-
[06.2023]
-
Our paper has been accepted for the Journal of Electrochemistry Society
-
Machine learning benchmarks for the classification of equivalent circuit models from solid-state electrochemical impedance spectra [Journal (Open access)] [arXiv] [GitHub]
-
-
-
[04.2023]
I received an award for the poster presentation at Advanced Battery Power Conference held in Aachen, Germany [link to news] -
[03.2023]
I gave four invited talks at the following institutions:-
Battery Modelling Webinar Series (BMWS) hosted by Carnegie Mellon University [Slide]
-
Machine Learning Group hosted by Prof. Motonobu Kanagawa, EURECOM, France [link] [Slide]
-
Machine and Human Intelligence research group hosted by Prof. Luigi Acerbi, University of Helsinki, Finland
-
Data Science and AI Research Group hosted by Dr. Festus Adedoyin, Bournemouth University, UK.
-
-
[09.2022]
Our paper has been accepted at NeurIPS 2022. -
-
Fast Bayesian inference with batch Bayesian quadrature via kernel recombination [Paper] [arXiv] [OpenReview] [GitHub]
-
-
[06.2022]
We organised the international battery data science hackathon event BatteryDEV 2022.[link]
Publications
Peer-reviewed conference publication (6 first author papers, 2 co-author paper, 4 under review)
-
M. Adachi*, W. Xu*, C.N. Jones M.A. Osborne (2024). "Principled Bayesian Optimization in Collaboration with Human Experts", Advances in Neural Information Processing Systems (NeurIPS Spotlight). *: equal contribution, acceptance rate 3% [arXiv] [GitHub]
-
M. Adachi*, S. Hayakawa*, M. Jørgensen, H. Oberhauser, M.A. Osborne (2022). "Fast Bayesian inference with batch Bayesian quadrature via kernel recombination", Advances in Neural Information Processing Systems (NeurIPS) 35, 16533 - 16547. *: equal contribution, acceptance rate 25.6%
[Paper] [arXiv] [OpenReview] [GitHub] [Video] -
M. Adachi, S. Hayakawa, M. Jørgensen, X. Wan, H. Oberhauser, M.A. Osborne (2023). "Adaptive Batch Sizes for Active Learning: A Probabilistic Numerics Approach", Artificial Intelligence and Statistics (AISTATS) 27 [arXiv][GitHub], acceptance rate 27.6%
-
M. Adachi, B. Planden, D.A. Howey, Mi.A. Osborne, S. Orbell, N. Ares, K. Muandet, S.L.Chau (2023). "Looping in the Human: Collaborative and Explainable Bayesian Optimization", Artificial Intelligence and Statistics (AISTATS) 27 [arXiv][GitHub], acceptance rate 27.6%
-
M. Adachi, Y. Kuhn, B. Horstmann, A. Latz, M. A. Osborne, D. A. Howey (2023). "Bayesian Model Selection of Lithium-Ion Battery Models via Bayesian Quadrature", The 22nd World Congress of the International Federation of Automatic Control (IFAC),
[Journal (Open access)][arXiv] [GitHub] -
J. Ziomek, M. Adachi, M.A. Osborne, "Bayesian Optimisation with Unknown Hyperparameters: Regret Bounds Logarithmically Closer to Optimal", Advances in Neural Information Processing Systems (NeurIPS), acceptance rate 25.8%
-
J. Ziomek, M. Adachi, M.A. Osborne, "Beyond Lengthscales: No-regret Bayesian Optimisation With Unknown Hyperparameters Of Any Type", [arXiv]
-
M. Adachi, S. Hayakawa, S. Hamid, M. Jørgensen, H. Oberhauser, M.A. Osborne (2023). "SOBER: Highly Parallel Bayesian Optimization and Bayesian Quadrature over Discrete and Mixed Spaces", [arXiv] [GitHub]
Peer-reviewed journal articles (4 first author papers, 5 co-author papers)
-
J. Schaeffer, P. Gasper, E. G. Tamayo, R. Gasper, M. Adachi, J. P. G. Cardona, S. M. Bedoya, A. Bhutani, A. Schiek, R. Goodall, R. Findeisen, R. D. Braatz, S. Engelke, (2023) "Machine learning benchmarks for the classification of equivalent circuit models from solid-state electrochemical impedance spectra." Journal of Electrochemical Society, 170, 060512
[Journal (Open access)] [arXiv] [GitHub] -
H. Yamahara, B. Feng, M. Seki, M. Adachi, et al., (2021). Flexoelectric Nanodomains in Rare-Earth Iron Garnet Thin Films under Strain Gradient, Communications Materials 2, 95
[Paper] -
W. Xiaohan, J. Billaud, I. Jerjen, F. Marone, Y. Ishihara, M. Adachi, Y. Adachi, C. Villevieille, Y. Kato (2019). Operando Visualization of Morphological Dynamics in All-Solid-State Batteries, Advanced Energy Materials 9, 1901547
[Paper] [ChemRxiv] -
M. Adachi., M. Seki, H. Yamahara, H. Nasu, H. Tabata. (2015). Long-term potentiation of magnonic synapses by photocontrolled spin current mimicked in reentrant spin-glass garnet ferrite Lu3Fe5-2xCoxSixO12 thin films. Applied Physics Express 8,4, 043002-1-4.
[Paper] -
M. Adachi., H. Matsui, M. Seki, H. Yamahara, H. Tabata. (2015). High-temperature terahertz absorption band in rare-earth gallium garnet. Physical Review B 91, 085118.
[Paper] -
H. Yamahara, M. Seki, M, Adachi, M. Takahashi, H. Nasu, K. Horiba, H. Kumigashira, H. Tabata. (2015). Spin-glass behaviors in carrier polarity controlled Fe3-xTixO4 semiconductor thin films. Applied Physics 118, 6, 063905
[Paper] -
M. Adachi., H. Yamahara, S. Kawabe, H. Matsui, H. Tabata. (2014). Strong optical reflection of rare-earth garnets in the terahertz regime by reststrahlen bands. Physical Review B 89, 205124.
[Paper] -
M. Seki, M. Takahashi, M, Adachi, H. Yamahara, H. Tabata. (2014). Fabrication and characterization of wüstite-based epitaxial thin films: p-type wide-gap oxide semiconductors composed of abundant elements. Applied Physics Letters 105, 112105
[Pape]
Conference Presentations
-
Oral Talks (4 international and 7 domestic conference)
-
Posters Presentation (3 international and 1 domestic conference)
Talks
Invited Talks
-
29.02.2024, LAM Research, CA, USA, hosted by Dr. Joe Lu.
-
24.03.2023, Machine Learning Group hosted by Prof. Motonobu Kanagawa, EURECOM, France
Machine Learning for Science via Fast Bayesian Optimisation and Quadrature [Slide] - 09.03.2023, Data Science and AI Research Group hosted by Dr. Festus Adedoyin, Bournemouth University, UK.
-
06.03.2023, Machine and Human Intelligence research group hosted by Prof. Luigi Acerbi, University of Helsinki, Finland.
-
08.02.2023, Battery Modelling Webinar Series (BMWS) hosted by Carnegie Mellon University, 08.02.2023 [Slide]
- 04.03.2022, Toyota Research institute
Oral & Poster presentations
-
26.04.2023, Advanced Battery Power Conference, Balancing accuracy and complexity when selecting battery models to fit data [Poster]
-
22.03.2023, ModVal 2023, Fast uncertainty prediction in digital twins of lithium-ion batteries via Bayesian quadrature [Poster]
-
30.12.2022, NeurIPS 2022, Fast Bayesian inference with Batch Bayesian Quadrature via Kernel Recombination [Slide]
-
14.03.2022, ModVal 2022, Bayesian Quadrature for Fast Parameter Estimation of a Lithium-ion Battery Model [Slide]
Working Experiences
Data Scientist
Toyota Motor Corporation
Data-driven design of next-generation batteries and motors for revolutionising electric vehicles:
intrapreneur based on intelligent experimental data analytics service
Machine learning analysis of manufacturing & experimental data (Python, R)
Deep learning model development (Pytorch)
Research on next-generation batteries, including materials discovery of new materials using machine learning
APRIL 2018 - PRESENT
Embedded Researcher
University of Cambridge
Belonging to two laboratories at the same time:
Machine Intelligence Laboratory, Department of Engineering (Host: Professor Roberto Cipolla)
Electron Microscope Laboratory, Department of Materials Science & Metallurgy (Host: Professor Caterina Ducati)
Collaborating with Professor Clare Grey
FEBRUARY 2020 - JANUARY 2021
Energy Scientist
Toyota Motor Corporation
Simulation-Based Design of Lithium-Ion Battery for Hybrid Vehicles (4th Prius)
Development of simulation software
Robust designing of commercial lithium ion batteries (HV, PHV, and EV)
Experimental skills of lithium ion batteries (fabrication & measurements)
APRIL 2015 - MARCH 2018
Education
OCTOBER 2021 - PRESENT
DPhil in Machine Learning
University of Oxford
Clarendon Scholar (the crème de la crème of the incoming cohort)
Thesis: Fast Bayesian Inference with Quadrature for Battery Control
Primary supervisor: Professor Michael Osborne,
Secondary supervisor: Professor David Howey
APRIL 2013 - MARCH 2015
MEng in Electronic Engineering
University of Tokyo
Summa cum laude (ranked 1st in the cohort of 3,074 MEng graduates)
Department of Electrical Engineering and Information Systems,
Thesis title: Neuromorphic computing using frustrated magnetic thin films
Awarded President’s Prize (the best master’s thesis in the cohort)
Superviser: Professor Hitoshi Tabata
Awards
Oxford Period (5 awards)
-
Poster Award, Advanced Battery Power Conference, 2023
-
NeurIPS 2022 Scholar Award, NeurIPS, 2022
-
Clarendon Scholarship, Clarendon Fund, 2021
-
Oxford Kobe Scholarship, University of Oxford, 2021
-
Toshizo Watanabe International Scholarship, The Watanabe Foundation, 2021
-
BCJA Scholarship, British Council, 2021
Toyota Period (5 awards)
-
Best Research Award, Toyota Engineering Society, 2019
-
Best Research Award, Toyota Engineering Society, 2018
-
Most Invented Prize, Toyota Motor Corporation, 2017
-
Outstanding Youth Commendation, Toyota City Council, 2016
-
Most Invented Prize, Toyota Motor Corporation, 2016
Univ. Tokyo Period (6 awards)
-
Young Researcher Award, 47th international conference on Solid-State Devices and Materials, 2015
-
President's Award, University of Tokyo, 2015
-
Dean's Award, Department of Engineering, University of Tokyo, 2015
-
Best Master's Thesis Award, EEIS University of Tokyo, 2015
-
Outstanding Student Completion Award, Institute of EICE society, 2015
-
JASSO Scholarship, JASSO, 2015
Patent
Patent Issued (5 Patents)
-
Electrolyte of non-polar solvent and bipolar salt LiBPh4 for lithium ion batteries, Issued Jun 9, 2021 Patent JP06895079
-
Fluoride ion conductor PbSnF2 coated negative electrode for five volt lithium-ion batteries, Issued Apr 21, 2021, EPP3547409, DE3547409, FR3547409, GB3547409, US3547409
-
high-power lithium-ion batteries using cyclopentyl methyl ether of electrolyte, Issued Nov 26, 2020, JP06799783
-
Mixed Zeta potential oxides coated cathode materials for high power lithium ion batteries, Issued Nov 4, 2020, EPP3547419, DE3547419, FR3547419, GB3547419, US3547419
-
Magnetic-phase-transitional porous metal-complex nanocoils for high-power lithium-ion batteries, Issued Jun 14, 2019, JP06536908
Patent Filed (8 Patents)
-
Algorithm to efficiently discover the desired materials, Aug 2021
-
Algorithm to support the chemical elemental composition inference from crystal structural spectra, Aug 2021
-
Algorithm to support the new material discovery from spectral data mining, Aug 2021
-
Automatic spectral identification of materials using hierarchical deep metric learning, Aug 2021
-
Automatic inspection of hue/texture anomaly for car leathers based on camera images, Mar 2021
-
Nano-porous micro-structured silicide alloy framework for high-capacity lithium-ion solid-state batteries, Filed Mar 23, 2021, 202100588JP00
-
Novel cathode material Bi2FeCoO3F6 for high-capacity fluorine-ion batteries, Filed Sep 6, 2019, 201904863JP00
-
Si - NiTi nano-composite sub-nanoparticles for high-capacity lithium-ion solid-state batteries, Filed Feb 7, 2018, 180892JP