革新知能統合研究センター

自然言語処理のトップカンファレンスの一つであるEMNLP2019において、AIPセンターから8本の論文が採択されました。EMNLP2019は、11月3日～7日に香港で開催されます。

[Website] https://www.emnlp-ijcnlp2019.org/
[Accepted Papers] https://www.emnlp-ijcnlp2019.org/program/accepted/

Long Papers

• Empirical Study of Transformer’s Attention Mechanism via the Lens of Kernel
  Yao-Hung Hubert Tsai, Shaojie Bai, Makoto Yamada (RIKEN AIP), Louis-Philippe Morency and Ruslan Salakhutdinov
• Learning Only from Relevant Keywords and Unlabeled Documents
  Nontawat Charoenphakdee (RIKEN AIP), Jongyeong Lee (RIKEN AIP), Yiping Jin, Dittaya Wanvarie and Masashi Sugiyama (RIKEN AIP)
• Select and Attend: Towards Controllable Content Selection in Text Generation
  Xiaoyu Shen (RIKEN AIP internship student), Jun Suzuki (RIKEN AIP), Kentaro Inui (RIKEN AIP), Hui Su, Dietrich Klakow and Satoshi Sekine (RIKEN AIP)

Short Papers

• An Empirical Study of Incorporating Pseudo Data into Grammatical Error Correction
  Shun Kiyono (RIKEN AIP), Jun Suzuki (RIKEN AIP), Masato Mita (RIKEN AIP), Tomoya Mizumoto and Kentaro Inui (RIKEN AIP)
  ※本研究では、自然言語処理の基盤技術の一つである文法誤り訂正において、ベンチマークデータ上で世界最高性能を達成しました。理研AIPセンターが所有するNVIDIA DGX-2を用いて行われました。
• Bridging the Defined and the Defining: Exploiting Implicit Lexical Semantic Relations in Definition Modeling
  Koki Washio (RIKEN AIP), Satoshi Sekine (RIKEN AIP) and Tsuneaki Kato
• Multi-Task Learning for Chemical Named Entity Recognition with Chemical Compound Paraphrase
  Taiki Watanabe (RIKEN AIP), Akihiro Tamura (RIKEN AIP), Takashi Ninomiya (RIKEN AIP), Takuya Makino (RIKEN AIP) and Tomoya Iwakura (RIKEN AIP)
• Transductive Learning of Neural Language Models for Syntactic and Semantic Analysis
  Hiroki Ouchi (RIKEN AIP), Jun Suzuki (RIKEN AIP) and Kentaro Inui (RIKEN AIP)

Demo Paper

• TEASPN: Framework and Protocol for Integrated Writing Assistance Environments
  Masato Hagiwara, Takumi Ito, Tatsuki Kuribayashi, Jun Suzuki (RIKEN AIP) and Kentaro
  Inui (RIKEN AIP)

関連研究室