Lis Kanashiro Pereira
カナシロ ペレイラ リズ
I’m a Project-Based Assistant Professor working on Natural Language Processing (NLP) at the Social Computing Laboratory at NAIST, Japan.
I'm currently working on representation learning methods for NLP.
Contact Info: x@is.naist.jp where x = kanashiro.lis
Address: Social Computing Lab., Division of Information Science, Graduate School of Science and Technology, Nara Institute of Science and Technology (NAIST). 8916-5, Takayama-cho, Ikoma, Nara 630-0192, Japan.
RECENT activities
**New**: Will be Area Chair for ACL and EMNLP 2023.
**New**: Gave a talk at the 「言語による時間生成」研究報告会, at Meikai University, March 21st, 2023, Chiba, Japan.
Presented a poster at the Chronogenesis: How the Mind Generates Time Meeting, held on March 5th and 6th, 2022, Hiroshima, Japan.
Gave a talk at the International Symposium on Chronogenesis: How the Mind Generates Time, held on November 23 (Wed) & 24 (Thu), 2022, at the Bankoku Shinryokan, Summit Hall, Okinawa, Japan.
Our paper Toward Building a General-Purpose Language Model for Understanding Temporal Commonsense has been accepted to the Student Research Workshop @ AACL 2022.
Our paper Effective Masked Language Modeling for Temporal Commonsense Reasoning has been accepted to SCIS&ISIS2022.
Gave a talk at AIST, Knowledge and Information Research Team.
Our paper OCHADAI at SemEval-2022 Task 2: Adversarial Training for Multilingual Idiomaticity Detection has been accepted to SEMEVAL 2022.
Our paper Attention is All you Need for Robust Temporal Reasoning has been accepted to LREC 2022.
Gave a talk at Microsoft Redmond Campus.
Our paper ALICE++: Adversarial Training for Robust and Effective Temporal Reasoning has been accepted to PACLIC 2021.
Our paper Multi-Layer Random Perturbation Training for Improving Model Generalization has been accepted to BlackBoxNLP@EMNLP 2021.
Our paper Towards a Language Model for Temporal Commonsense Reasoning has been accepted to the RANLP 2021 Student Workshop.
Our paper OCHADAI-KYODAI at SemEval-2021 Task 1: Enhancing Model Generalization and Robustness for Lexical Complexity Prediction won an Honorable Mention at SemEval-2021! --> https://semeval.github.io/SemEval2021/awards
Our recent paper Targeted Adversarial Training for Natural Language Understanding has been accepted to NAACL-2021. #1 on the XNLI dataset
Our recent paper Posterior Differential Regularization with f-divergence for Improving Model Robustness has been accepted to NAACL-2021.
Our recent paper OCHADAI-KYODAI at SemEval-2021 Task 1: Enhancing Model Generalization and Robustness for Lexical Complexity Prediction has been accepted to be presented at SemEval 2021.
Our recent paper OCHADAI at SMM4H-2021 Task 5: Classifying self-reporting tweets on potential cases of COVID-19 by ensembling pre-trained language models has been accepted to be presented at SMM4H 2021@NAACL.
Our system ranked among the top-10 systems on both subtasks of SemEval-2021 Task 1 (Lexical Complexity Prediction).
Our recent paper Posterior Differential Regularization with f-divergence for Improving Model Robustness has been uploaded to arxiv.
Our model (ALICE) ranked 1st on the Allen Institute for AI's Leaderboard on the Temporal Commonsense Comprehension Task. (As of June 11th, 2020), outperforming the T5 model.
Our paper "Adversarial Training for Commonsense Inference" got accepted at the ACL 2020 Workshop on Representation Learning for NLP (RepL4NLP-2020).
Our model (ALICE) ranked 2nd on the Allen Institute for AI's Leaderboard on the SciTail dataset (Natural Language Inference dataset). (As of May 15th, 2020)
Our model (ALICE) ranked 1st on the Allen Institute for AI's Leaderboard on the CosmosQA dataset (Machine Reading Comprehension with Contextual Commonsense Reasoning). (As of March 23rd, 2020)
We ranked 1st on the Allen Institute for AI's Leaderboard on the Temporal Commonsense Comprehension Task. (As of February 27th, 2020)
Journal Reviewer: CALICO Journal
Conference Paper Reviewer: ACL, EMNLP, NAACL, AAAI, WSDM, SIGIR, AACL, BEA, BlackBoxNLP, SustaiNLP, GenBench
Program Chair: ACL 2023 (Machine Learning for NLP track), EMNLP 2023
Education
Doctor of Engineering (2013- 2016)
Computational Linguistics Laboratory, Nara Institute of Science and Technology, Japan.
Supervisor: Yuji Matsumoto.
Master of Engineering (2011- 2013)
Computational Linguistics Laboratory, Nara Institute of Science and Technology, Japan.
Supervisor: Yuji Matsumoto.
B.S. Computer Science (2003-2007)
Universidade Federal do Pará, Belém, Brazil.
Professional Experience
Project-Based Assistant Professor (Apr. 2023 - Present)
NAIST, Japan.
Project-Based Lecturer (Apr. 2021- Mar. 2023)
Ochanomizu University, Japan.
Project-Based Assistant Professor (Dec. 2019- Mar. 2021)
Ochanomizu University, Japan.
NLP Engineer (2019)
A.I. Squared Inc., Japan.
NLP Engineer (2017-2019)
Weathernews Inc., Japan.
Postdoctoral Fellow (2017)
City University of Hong Kong, Hong Kong.
Supervisor: John Lee
Information Technology Analyst, Federal Government Employee (2008- 2010)
Universidade Federal do Pará (UFPA), Information Technology and Communication Center (CTIC), Pará, Brazil
TALKS
Microsoft Redmond Campus, January 2022. Title: Robust and Generalizable Language Model Fine-Tuning.
AIST, Japan. May, 2022. Title: Robust and Generalizable Language Model Fine-Tuning.
International Symposium on Chronogenesis: How the Mind Generates Time, November 23 (Wed) & 24 (Thu), 2022 , Okinawa, Japan. Title: Adversarial Training for Robust Temporal Reasoning in Natural Language Processing.
「言語による時間生成」研究報告会, at Meikai University, March 21st, 2023, Chiba, Japan. Title: Robust Language Models for Japanese Temporal Reasoning.
Publications
船曳 日佳里, 木村 麻友子, Lis Kanashiro Pereira, 浅原 正幸, Fei Cheng，越智綾子，小林一郎. マルチタスク学習を用いた時間を認識する汎用言語モデルの構築. JSAI 2023. (To appear)
木村麻友子，Lis Kanashiro Pereira，浅原正幸，Fei Cheng，越智綾子，小林一郎．「時間関係タスクを対象にしたマルチタスク学習におけるデータの親和性の解析」．言語処理学会第 29 回年次大会，沖縄コンベンションセンター，沖縄，2023年3月.
船曳日佳里，Lis Kanashiro Pereira，木村麻友子，浅原正幸，Fei Cheng，越智綾子，小林一郎．「日本語の時間的常識を理解する言語モデルの構築を目的としたマルチタスク学習における検証」．言語処理学会第 29 回年次大会，沖縄コンベンションセンター，沖縄，2023年3月．
Mayuko Kimura, Lis Kanashiro Pereira and Ichiro Kobayashi. Toward Building a General-Purpose Language Model for Understanding Temporal Commonsense. Student Research Workshop @ AACL 2022.
Mayuko Kimura, Lis Kanashiro Pereira and Ichiro Kobayashi. Effective Masked Language Modeling for Temporal Commonsense Reasoning. SCIS&ISIS2022.
Lis Kanashiro Pereira, Kevin Duh, Fei Cheng, Masayuki Asahara, Ichiro Kobayashi . Attention-Focused Adversarial Training for Robust Temporal Reasoning. LREC 2022.
Lis Kanashiro Pereira, Ichiro Kobayashi. OCHADAI at SemEval-2022 Task 2: Adversarial Training for Multilingual Idiomaticity Detection. SemEval 2022.
Lis Pereira, Fei Cheng, Masayuki Asahara, and Ichiro Kobayashi. ALICE++: Adversarial Training for Robust and Effective Temporal Reasoning. PACLIC 2021.
Chenjing Geng, Fei Cheng, Masayuki Asahara, Lis Kanashiro Pereira, and Ichiro Kobayashi. Dependency Enhanced Contextual Representations for Japanese Temporal Relation Classification. PACLIC 2021.
Lis Pereira, Yuki Taya, , and Ichiro Kobayashi. Multi-Layer Random Perturbation Training for Improving Model Generalization. BlackBoxNLP@EMNLP 2021. (to appear)
Mayuko Kimura, Lis Kanashiro Pereira and Ichiro Kobayashi. Towards a Language Model for Temporal Commonsense Reasoning. Recent Advances in Natural Language Processing (RANLP 2021) Student Workshop.
Lis Pereira*, Xiaodong Liu*, Hao Cheng, Hoifung Poon, Jianfeng Gao and Ichiro Kobayashi. Targeted Adversarial Training for Natural Language Understanding. NAACL-2021 (*equal contribution). [pdf] #1 on the XNLI dataset
Hao Cheng, Xiaodong Liu, Lis Pereira, Yaoliang Yu, Jianfeng Gao. Posterior Differential Regularization with f-divergence for Improving Model Robustness. NAACL-2021. [pdf]
Yuki Taya, Lis Pereira, Fei Cheng, Ichiro Kobayashi. OCHADAI-KYODAI at SemEval-2021 Task 1: Enhancing Model Generalization and Robustness for Lexical Complexity Prediction. SemEval 2021@ACL. (honorable mention) [pdf]
Ying Luo, Lis Pereira, Ichiro Kobayashi. OCHADAI at SMM4H-2021 Task 5: Classifying self-reporting tweets on potential cases of COVID-19 by ensembling pre-trained language models. SMM4H 2021@NAACL.
Murayama, Y., Pereira, L. and Kobayashi, I. Dialogue over Context and Structured Knowledge using a Neural Network Model with External Memories. KNLP@AACL-IJCNLP 2020. [pdf]
Pereira, L., Liu, X., Cheng, F., Asahara, M. and Kobayashi, I. 2020. Adversarial Training for Commonsense Inference. ACL 2020 Workshop on Representation Learning for NLP (Rep4NLP@ACL2020). [pdf] #1 on the CosmosQA and on the MC-TACO datasets
Pereira, L., Liu, X., and Lee, J. 2017. Lexical Simplification using Deep Semantic Structured Models. In Proceedings of the Eighth International Joint Conference on Natural Language Processing (Volume 2: Short Papers), pp. 430-435, Taipei, Taiwan.
Lis Pereira and Yuji Matsumoto. Leveraging a Learner Corpus for Automated Collocation Suggestion for Learners of Japanese as a Second Language. CALICO Journal. vol 33.3 2016 311–333, doi : 10.1558/cj.v33i3.26444
Lis Pereira and Yuji Matsumoto. 2015. Collocational Aid for Learners of Japanese as a Second Language. In Proceedings of the ACL 2015 Workshop on Natural Language Processing Techniques for Educational Applications (NLP-TEA-2), Beijing, China.
Lis W.K. Pereira, Elga Strafella, Kevin Duh and Yuji Matsumoto. 2014. Identifying Collocations using Cross-lingual Association Measures. In Proceedings of the EACL 2014 Workshop on Multiword Expressions, Gothenburg, Sweden.
Lis W.K. Pereira, Elga Strafella and Yuji Matsumoto. 2014. Collocation or Free Combination? – Applying Machine Translation Techniques to identify collocations in Japanese. Proceedings of the 9th Language Resources and Evaluation Conference (LREC 2014), Reykjavik, Iceland.
Lis W.K. Pereira, Erlyn Manguilimotan and Yuji Matsumoto. 2013. Automated Collocation Suggestion for Japanese Second Language Learners. In Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics (ACL), Student Research Workshop, pp.52-58, Sofia, Bulgaria.
Keisuke Sakaguchi, Yuta Hayashibe, Shuhei Kondo, Lis Kanashiro, Tomoya Mizumoto, Mamoru Komachi and Yuji Matsumoto. NAIST at the HOO 2012 Shared Task. The Seventh Workshop on Building Educational Applications Using NLP, pp.281-288, Montreal, Canada, 2012 June 7th.
Domestic Conferences
木村 麻友子，Kanashiro Pereira Lis，浅原 正幸，Cheng Fei，越智 綾子，小林 一郎. 「時間的常識理解へ向けた言語モデル構築への取り組み」. 人工知能学会全国大会（第36回），国立京都国際会館，京都，2022年6月．
船曳 日佳里，木村 麻友子，Kanashiro Pereira Lis，小林 一郎.「時間的常識を認識する日本語汎用言語モデルの構築への取り組み」. 人工知能学会全国大会（第36回），国立京都国際会館，京都，2022年6月．
木村麻友子, Lis Kanashiro Pereira (お茶大), 浅原正幸 (国語研), Fei Cheng (京大), 越智綾子 (国語研), 小林一郎 (お茶大). 時間的常識理解へ向けた効果的なマスク言語モデルの検証. 言語処理学会第28回年次大会(NLP2022)
Kimura, M., Pereira, L. and Kobayashi, I. Towards a Temporal Commonsense Aware Language Model. In Proceedings of the Twenty-Seventh Annual Meeting of the Association for Natural Language Processing (NLP-2021). (In Japanese)
深層強化学習モデルの内部挙動の言語化による制御手法構築へ向けて
圓田彩乃、Lis Kanashiro Pereira、小林一郎, Domestic, 2021.06, 第35回人工知能学会全国大会, オンライン
特性を顕在化する言語の意味を反映した画像生成
渡邊清子、Lis Kanashiro Pereira、小林一郎, Domestic, 2021.06, 第35回人工知能学会全国大会, オンライン
Lis W.K. Pereira, Erlyn Manguilimotan and Yuji Matsumoto. 2013. Data Coverage vs. Data Size: A comparison of two large-scale corpora in Collocation Suggestion for Japanese Second Language Learners. In Proceedings of the Nineteenth Annual Meeting of the Association for Natural Language Processing (NLP-2013), pp.74-76, Nagoya, Japan, March 2013.
Lis W.K.Pereira, Erlyn Manguilimotan and Yuji Matsumoto. 2013. Collocation Suggestion for Japanese Second Language Learners, 情報処理学会研究報告 第２１０回自然言語処理研究、Vol.2013-NL-210、No.３、pp.1-5, January 2013.
Pereira, Lis W.K., Chagas, Larissa F., Souza, Jñane (2009) , Melhoria de Processo de Software no CTIC-UFPa: Um relato de experiência, III Workshop de Tecnologia de Informação das IFES, Brazil.
Mota, Marcelle; Pereira, Lis W.K. & Favero, Eloi (2008), JavaTool, uma ferramenta para ensino de programação, XXVIII Congress of the Brazilian Computer Society (SBC 2008), Brazil.