Atsuki Sato

ust1-25 → eeic2021 → ice2023

[松井研 / Matsui Lab @utokyo_bunny]

Ryotaro's paper below won the Best Paper Award at the SIGIR Workshop SynthIR! #SIGIR2026

[松井研 / Matsui Lab @utokyo_bunny]

Atsuki's paper has been accepted to #TMLR! We propose a framework for automatically constructing a sorting algorithm from a given (learned) index. Given index construction O(nC(n)) and rank query O(Q(n)), it yields sorting in O(nC(n)+nQ(n)). https://t.co/Es3tbzB5HI

[松井研 / Matsui Lab @utokyo_bunny]

We will present our work with Atsuki and Martin at #SIGMOD2026 🇮🇳 We provide a theoretical foundation for poisoning attacks against learned indexes, where an adversary inserts malicious data to degrade performance😈 https://t.co/CgK96z2OLT

[Atsuki Sato @chikkun282]

@not_iii_but_jjj こそこそと監視すな

[Atsuki Sato @chikkun282]

@not_iii_but_jjj 言うわけがないし、再会な

[松井研 / Matsui Lab @utokyo_bunny]

The paper on Boosting by Haruki, who was an undergraduate student in the Matsui Lab last year, has been accepted to #ICML2026 🙌🎉🇰🇷

[Liner @search_liner]

Liner is partnering with @spoticlr at #ICLR2026 — supporting Best Paper and Travel Awards for LLM research.

And to celebrate, we're giving away:
✈️ Round-trip flights + hotel to #ICML2026 in Seoul
🎁 $300 Liner Credits

Follow @search_liner + repost to enter by 4/27.

Liner is built for research workflows. Find papers, verify sources, and write with citations in one place.

See you in 🇧🇷 and ����🇷!

@iclr_conf @icmlconf

[松井研 / Matsui Lab @utokyo_bunny]

Thrilled to share that our paper with Atsuki (PhD student) and Martin (IT Copenhagen) has been accepted to #SIGMOD2026 🎉

[松井研 / Matsui Lab @utokyo_bunny]

Atsuki's paper has been accepted to TMLR with a Featured Certification! We present a learned sorting algorithm that sorts data in O(nloglogn) when the data distribution satisfies certain conditions. #TMLR https://t.co/FdiFX9xl6m

[松井研 / Matsui Lab @utokyo_bunny]

D1佐藤君の論文がTMLRにFeatured Certificationで採択されました！データ分布が条件に従うときにデータをO(nloglogn)でソートする、学習型ソーティングです #TMLR https://t.co/FdiFX9xl6m

[松井研 / Matsui Lab @utokyo_bunny]

Fuma's paper on fast generation of surface filling curves has been accepted to SIGGRAPH Asia 2025 Technical Communications!🐇

[松井研 / Matsui Lab @utokyo_bunny]

M2日髙君の、3Dモデルの表面を充填する曲線の高速な生成に関する論文が、SIGGRAPH Asia 2025 Technical Communicationsに採択されました！🐇

[松井研 / Matsui Lab @utokyo_bunny]

Two papers have been accepted to the NeurIPS 2025 Workshop on ML for Systems! #MLforSystems #NeurIPS2025

(1) Learned Count-Min Sketch by Kyosuke and Atsuki

(2) Evaluation of ANN Deletion by Tomohiro and Prof. Amagata @Daichi_Amagata (UOsaka)

[松井研 / Matsui Lab @utokyo_bunny]

NeurIPS 2025 WS on ML for Systemsに論文が���本採択されました！ #MLforSystems #NeurIPS2025

(1) 去年卒論の西下君、D1佐藤君による学習型カウントミンスケッチ

(2) M1の山下君、阪大 @Daichi_Amagata 先生によるANN削除方式の評価

[Atsuki Sato @chikkun282]

Rejected #neurips2025

[松井研 / Matsui Lab @utokyo_bunny]

We have released the code for LotusFilter presented at CVPR25! It is a super-fast post-processing module that diversifies Faiss search results. You can try it out with `pip install lotf`. Internally, it uses C++17 and Boost Unordered with nanobind. https://t.co/0s5vTyNmV9

[松井研 / Matsui Lab @utokyo_bunny]

We'll host the Workshop on Vector Databases #VecDB tomorrow at #ICML2025🙌. We warmly invite you to join us! By registering, you can watch the orals/keynotes remotely.
🕑: July 18, 8:30am-5:30pm
🚪: West Meeting Room 208
��: https://t.co/6uGh56xhhH
📺: https://t.co/Xu9xKTd2Of

[松井研 / Matsui Lab @utokyo_bunny]

I'll present my poster paper at #CVPR2025 on June 15! I propose an extremely fast post-processing module for diverse nearest-neighbor searches🚀

Y. Matsui, "LotusFilter: Fast Diverse Nearest Neighbor Search via a Learned Cutoff Table"

https://t.co/QDak907mfZ

[東京大学 | UTokyo @UTokyo_News]

[プレスリリース] 非常に高速で多様な情報検索が可能に！――生成AIやECサイト検索結果向上から地震の波形探索への応用まで―――情報理工学系研究科
https://t.co/Ahx37pyCZV

[Atsuyuki Miyai @UTokyo @AtsuMiyaiAM]

🧙‍♂️ Imagine web agents that don’t just browse but handle your tedious digital chores!

📣 Our team developed WebChoreArena
- 532 human-curated tasks, crafted over 300+ hours
- Tests agents on massive information memorization, mathematical reasoning, and long-term memory
- built on WebArena with full reproducibility

📊 Even Gemini 2.5 Pro shows substantial room for improvement, highlighting a critical challenge for the next LLMs-based web agents!

🌐 https://t.co/m9vFtdRB4p
📕 https://t.co/zfXvs6MCtF
🧑‍💻 https://t.co/icPFtopTLL