当研究室の修士2年の村瀬君と藤井君が11月下旬から12月中旬まで、豊田工業大学シカゴ校(TTIC: Toyota Technological Institute at Chicago)に協定留学します。通常は3ヶ月間の協定留学期間がありますが、コロナ禍の影響により短縮となりました。来年からは通常の協定留学に戻ることを願っています。

BioCreative VII presentations

Our submissions have been accepted for poster and short presentations at the BioCreative VII Workshop.

DrugProt:Text mining drug/chemical-protein interactions (Track 1)

Naoki Iinuma, Masaki Asada, Makoto Miwa, Yutaka Sasaki, TTI-COIN at BioCreative VII Track 1, BioCreative VII Workshop, November, 2021. (pdf)

Selected poster flash talks

Tomoki Tsujimura, Ryuki Ida, Isanori Oiwa, Makoto Miwa, and Yutaka Sasaki, TTI-COIN at BioCreative VII Track 2, BioCreative VII Workshop, November, 2021. (pdf)

GAN-based Shape Design for Die Forging

New preprint

Hayato Futase, Tomoki Tsujimura, Tetsuya Kajimoto, Hajime Kawarazaki, Toshiyuki Suzuki, Makoto Miwa, Yutaka Sasaki,Physical Context and Timing Aware Sequence Generating GANs, arXiv:2110.04077, 2021.


This paper proposes a new neural model of Generative Adversarial Networks (GANs) that is suitable for designing the intermediate shapes for die forging. Because the shape of heated metal gradually changes during the press, GANs should be aware of the timing and physical phenomena. In this respect, we proposed Physical Context and Timing aware sequence generating GANs (PCTGAN) that generates images in a sequence, considering the time sequence and physical quantities.

This new GAN is partly patented (U.S. Patent No: 10,970,601).

Journal paper accepted

Our paper has been accepted for publication on Frontiers in Research Metrics and Analytics, 2021.

This paper proposes a new Knowledge Graph dataset that includes not only heterogeneous node relations but also textual descriptions. The dataset targets pharmaceutical areas but it can contribute to the Knowledge Graph research community by providing evidences in the usefulness of textural information integrated in a knowledge graph.

Masaki Asada, Gunasekaran Nallappan, Makoto Miwa and Yutaka Sasaki, Representing a Heterogeneous Pharmaceutical Knowledge-Graph with Textual Information, Front. Res. Metr. Anal., doi: 10.3389/frma.2021.670206, 2021.


Paper accepted for ACL2021 Findings

Our conference paper by PhD student Makino has been accepted for Findins of ACL.

D1の牧野君の論文がFindins of ACL 2021に採録されました.

Findings はACLの新しい施策で,発表のない論文集としてACLの本会議論文とともに公開されます.ACLの厳しい査読を経て採録されていますので査読付きの業績となります.実際,査読のスコアは4.0, 3.5, 2.5でしたのでACL本会議論文と遜色のない内容だと思っています.私が査読者として査読する場合,4.0は滅多に付けないので,査読者のひとりがかなりの自信をもってサポートしてくれている点は喜んでいいと思っています.

2021年言語処理学会年次大会 スポンサー賞受賞


◎C8:知識獲得・情報抽出(5)   3月18日(木) 11:00-12:30 22会議室/Zoom
C8-3 辺編集による文書レベルの関係グラフ構築
○牧野晃平, 三輪誠, 佐々木裕 (豊田工大)