Data is a source of knowledge, and AI is the most powerful engine for unlocking it. AI has enabled black hole visualization and guerrilla heavy rain prediction. AI-powered services have attracted enormous wealth to tech giants such as GAMMA (Google, Apple, Microsoft, Meta, Amazon) and BATH (Baidu, Alibaba, Tencent, Huawei). In order to produce such results, it is essential to have systems technology that can process, train on, and serve huge amounts of data. We are creating new technologies for such AI systems, from large-scale data infrastructure to training and inference platforms to edge AI execution environments. These AI systems are the driving force for creating new science and services. Since AI systems are sandwiched between applications and hardware, we collaborate with researchers and practitioners in a wide range of fields, from ML to device design. Would you like to create a wonderful future with us?
Recent News
■ 2026.3.18 Yamato Kawaue won a silver medal at The 2026 ICPC Asia Pacific Championship held in Taiwan, finishing 12th overall. He has qualified for the ICPC World Finals to be held in November. Congratulations.
■ 2026.3.5 Aoi Kida received the Student Presentation Award at the 18th Forum on Data Engineering and Information Management (DEIM 2026).
■ 2026.3.4 The following paper has been accepted at APDCM, a workshop co-located with IPDPS.
Title: Immediate Deterministic Leader Election (IDLE) Algorithm for Reducing Downtime in Distributed Systems
Author: Masahiro Tanaka, Hideyuki KawashimaVenue: APDCM workshop in conjunction with IPDPS
■ 2026.3.1 The following paper has been accepted at CLEO. This work computes the Transformer Attention mechanism using laser beams (optics) instead of GPUs.
Title: Real-AI Workload Demonstration Using a Simple Photonic Dot-Product System: BERT Attention and AI Exposure Score
Author: Tatsuhiro Nakamori, Junnosuke Kokubu, Sayaka Tayama, Mitsumasa Nakajima, Hidetaka Nishi, Shinji Matsuo, Takasumi Tanabe, Hideyuki Kawashima
■ 2026.2.24 Harunari Takata's paper has been accepted at ACM SIGMOD (CORE Rank = A*). This is a joint work with the University of Toronto (Kawashima is not a co-author).
Title: Epoch-based Optimistic Concurrency Control in Geo-replicated Databases
Author: Yunhao Mao, Harunari Takata, Michail Bachras, Yuqiu Zhang, Shiquan Zhang, Gengrui Zhang, Hans-Arno Jacobson
Venue: ACM SIGMOD (PACMMOD)
■ 2026.2.20 A student has been accepted into the Virginia Tech Computer Science and Applications PhD program.
■ 2025.12.26 Two students have been selected as JSPS DC1 Research Fellows. Congratulations. All PhD students in our lab next semester (3 in total) will be DC1 fellows. JSPS DC1 (JSPS Research Fellowship for Young Scientists DC1) is a prestigious fellowship for doctoral students in Japan. Students apply during their M2 year, and if selected, receive a monthly stipend of ¥200,000 and up to ¥1,500,000 per year in research funding for up to 3 years starting from D1. It is widely regarded as a gateway for young researchers. The acceptance rate was 11%.
■ 2025.12.15 The following journal paper has been accepted. The first author is Tatsuhiro Nakamori (PhD student).
Title: Libra: One-Shot Parameter Sensitivity Estimation for Transfer Learning in Database Performance Prediction
Author: Tatsuhiro Nakamori, Hideyuki Kawashima
Venue: PVLDB Vol. 19, to appear at 52nd International Conference on Very Large Data Bases (VLDB'26)
■ 2025.10.13 Tatsuhiro Nakamori (PhD student) gave a poster presentation entitled "Libra: One-Shot Parameter Sensitivity Estimation for Transfer Learning in Database Performance Prediction" at SOSP 2025 (The 31st Symposium on Operating Systems Principles)
■ 2025.9.20 Tatsuhiro Nakamori (M2) received the Aiso Award (相磯賞) for the best master's thesis in the Class of 2026, Graduate School of Media and Governance,
■ 2025.9.4 Hideyuki Kawashima (PI) gave a talk about a new transaction processing method, Oze, at the 51st International Conference on Very Large Data Bases (VLDB) which is CORE rank = A*.
■ 2024.6.6 Two groups from our lab have been selected for the MITOU IT program. Congratulation. In 2025, there were 227 applications, and 21 were selected.
1. Foreign language learning application using subtitles for anime and dramas (Rika Shou)
2. Development of high-performance, fault-tolerant MySQL (Yusuke Miyazaki, Tatsuhiro Nakamori, Haowen Li
■ 2024.5.14 A student has been accepted into UBC（University of British Columbia）PhD Track.
■ 2025.4.16 The following paper has been accepted.
Title: Oze: Decentralized Graph-based Concurrency Control for Long-running Update Transactions
Author: Jun Nemoto, Takashi Kambayashi, Takashi Hoshino, Hideyuki Kawashim
Venue: VLDB'25 (PVLDBv18) (core rank = A*)
■ 2025.3.20 The following paper has been accepted.
Title: Campus: Making Cluster-based Vector Index Scalable with Transactions
Author: Masaru Uchida, HIdeyuki Kawashima
1st International Workshop on Coupling of Large Language Models with Vector Data Management (LLM+Vector Data) in conjunction with ICDE'25
■ 2025.3.2 The following paper has been accepted.
Title: Two-Step Quantum Search Algorithm for Solving Traveling Salesman Problems
■ 2025.2.21 We received the Best Student Poster Award at HPCAsia'25.
Title: Gato: Optimizing Multi-Version Deterministic Concurrency Control with Dynamic Partitioning
Author: Jiyeon Lim, Haowen Li, and Hideyuki Kawashima
Venue: HPCAsia'25.
■ 2024.1.27 The following paper has been accepted.
Title: Fairly Decentralizing a Hybrid Concurrency Control Protocol for Real-time Database Systems
Author: Tung Nguyen and Hideyuki Kawashima
Journal: Concurrency and Computation: Practice and Experience (CCPE)
■ 2024.12.11 The following paper has been accepted.
Title: Qsync: Extending Simplified SMR Protocol with Partial Network Partition Tolerance
Author: Aoi Kida and Hideyuki Kawashima
■ 2024.12.2 We received the Outstanding Paper Award at CANDAR symposium.
Title: Serval: A Wait-free Multi-version Deterministic Concurrency Control Scheme
Author: Haowen Li, Rina Onishi and Hideyuki Kawashima
Venue: The Twelfth International Symposium on Computing and Networking (CANDAR 2024)
■ 2024.9.27 We received the Best Poster Award at IEEE Cluster.
Title: Cheetah: An Efficient Deterministic Concurrency Control Scheme with Non-visible Write Elimination and Re-designed Garbage Collection
Author: Haowen Li, Rina Onishi and Hideyuki Kawashima
Venue: The IEEE International Conference on Cluster Computing (Cluster'24)
■ 2024.7 Sho Nakazono's paepr has been accepted for eScience (The 20th IEEE International Conference on e-Science)
Title: Griffin: Fast Transactional Database Index with Hash and B+-Tree
Author: Sho Nakazono, Yutaro Bessho, Hideyuki Kawashima and Tatsuhiro Nakamori
Venue: IEEE eScience
■ 2024.7 Kodai Doki's paper has been accepted for APSys (15th ACM SIGOPS Asia-Pacific Workshop on Systems)
Title: WoundDie: Concurrency Control Protocol with Lightweight Priority Control
Authors: Kodai Doki, Takashi Hoshino, Hideyuki Kawashima
Venue: APSys
■ 2024.6 A lab member (undergraduate student) has been accepted into the doctoral program at OIST (Okinawa Institute of Science and Technology Graduate University).
■ 2024.6 Masahide Fukuyama's paper has been accepted for a Journal (International Journal of Networking and Computing)
Masahide Fukuyama, Masahiro Tanaka, Ryota Ogino, Hideyuki Kawashima. eSilo: A Secure Transaction Processing System with SGX. IJNC.
■ 2024.6 Fukuyama has been certified as a IPA Mitou super creator.
https://www.ipa.go.jp/jinzai/mitou/it/2023/m42obm000000i86f-att/it2023_sc_booklet.pdf
■ 2024.6 Natsuki Hamada's paper has been accepted for a journal (IPSJ ACS).
Natsuki Hamada, Kazuhiro Saito, Hideyuki Kawashima. Qobra : Accelerating Transactional Serializability Verifier by Quantum Annealing. IPSJ Transactions on Advanced Computing Systems. To appear.
■ 2024.5 Aoi Kida gave a talk at an international workshop APDCM (IPDPS workshop)
Aoi Kida and Hideyuki Kawashima. Accelerating BFT Database with Transaction Reconstruction. 26th Workshop on Advances in Parallel and Distributed Computational Models (APDCM)
Past News
2020.12.13 SFCニュース：川島英之環境情報学部准教授らの研究成果が国際論文誌Proceedings of the VLDB Endowment (PVLDB)で出版
Y. Nakamura, H. Kawashima, O. Tatebe: “Integrating TicToc with Parallel Logging,” CANDAR workshops, 2018. Best Paper Award.
K. Hiraga, O. Tatebe, H. Kawashima: “PPMDS: A Distributed Metadata Server based on Nonblocking Transactions,” IEEE SNAMS workshops, 2018. Best Paper Award.
大黒, 川島, 建部: “Skew を考慮したMapReduce Shuffle 向け集団通信の設計と評価,” xSIG, 2018. Best M2 Student Award.
渡辺, 川島, 建部: “Masstree の一括構築法,” xSIG, 2018. Outstanding Significance Award.
田辺, 川島, 建部: “並行性制御法Cicada と並列ログ先行書き込み法P-WAL の結合,” xSIG, 2018. Outstanding B4 Student Award.
村田，川島，建部: “RDMA の適用によるRAMP トランザクション処理の高速化,” 情報処理学会論文誌データベース, 2017. 2018年度山下記念研究賞（川島が受賞）