Research seminar (2021)

毎週メンバー全員が参加し,研究について議論を行うミーティングです.

This is a weekly meeting where all members attend and discuss the research in the laboratory.

  • 日時 (Time slot):
    • 1Q: 12:30~ (Thu)
    • 2Q: 12:30~ (Thu)
  • 発表要領 (Presentation format):
    • (ひとりあたり)発表15 ~ 20分 および 議論10 ~ 15分,合計30分を目安にしてください.
    • 15 - 20 min for a presentation and 10 - 15 min for discussion, total 30 min.
    • Tipsの発表は5分程度(長くても質疑含めて15分くらい)が望ましいです
    • 5 mins for Tips presentation, upto 15 mins including discussion.
  • 言語 (Language used for presentation):
    • トークとスライドのうち,少なくともどちらか一方は英語で用意してください.
      学部生は例外的に,本人が希望すれば,トークとスライドの両方を日本語にすることができます.
    • Please prepare, between talk and slide, at least one of them must be English.
      Undergraduate students are exceptional: both talk and slide can be Japanese if they prefer it.
  • 座長 (Chair)
    • B/Mの学生は持ち回りで座長を担当します.座長の仕事を参考にして,円滑に議論が進むように心がけてください.
    • B/M students are assigned as a chair of the seminar. Please contribute to active discussion refering to Chair’s Job.

Planned Seminars

2022-03-10(Thu) 12:30~

  • NLP rehearsal
    • Ma: CNNによるNERとRE
    • Hiraoka(ポスター枠でした)
    • Yoshikawa 確信度を考慮した言語モデルの関係知識評価
    • Niwa: 近傍の事例を用いた非自己回帰生成

Past Seminars

2021-04-07(Wed) 12:45~

  • Agenda
    • seminar schedule(1Q)
    • assign seat
  • self-introductions by new members
    • Kaneko-san: self-introduction.pdf
    • Yoshikawa-san self-introduction.pdf
    • Vijay-san: self-introduction.pdf
    • Toyama-san: self-introduction.pdf
    • Ishikawa-san: self-introduction.pdf
    • Maeda-san pdf
    • Loem-san: self-introduction.pdf
    • Konishi-san
  • (FYI) self-introduction in previous years
    • 2020, 2019, 2018

2021-04-15(Thu) 12:30~

  • Kick-off
    • okazaki 研究室キックオフ(日本語) (English)
  • Agenda
    • New roles

2021-04-22(Thu) 12:30~

  • seminar
    • Hiraoka: Compositional Word Replacement
    • Wiem Behavioral Testing of Knowledge Graph Embedding Models
  • tips
    • Takase: What to Do in Lab
  • agenda:
    • Laboratory guidance (Apr 17)

2021-04-29(Thu) 12:30~

  • natinal holiday but lecture day.(祝日だが授業あり)
  • seminar
    • Mizuki: Learning Hierarchical Sense Code Representation for knowledge-based Word Sense Disambiguation
    • Kuo: Causality Relation Extraction
  • tips
    • Hiraoka: Keep Your Research Diary

2021-05-06(Thu) 12:30~

  • seminar
    • Sangwhan Slides

    • Ueki hallucination detection

  • tips
    • Niwa: How to manage information

2021-05-13(Thu) 12:30~

  • seminar
    • Iida slides
    • Koyama artificial error generation using various rules for neural grammatical error correction
  • tips
    • Muraoka: What I have learned in a company

    ### 2021-05-20(Thu) 12:30~

  • seminar
    • Muraoka: Evaluation of OSCAR on our IPP task
    • Taniguchi Advertisement generation research plans
  • tips
    • Mizuki: Two tips for kickstarting your research project

2021-05-27(Thu) 12:30~

  • agenda
    • seminar schedule of 2Q (move paper reading to Monday or Friday?)
  • seminar
    • Liu: Logic2text

    • Maruyama

  • tips
    • Yang VS code SSH

2021-06-03(Thu) 12:30~

  • seminar
    • Kaneko: Bias in distributed representation
    • Yoshikawa Research Plan: Estimating Confidence of Language Models’ Knowledge
  • tips
    • Iida pytorch_gpu_parallel

2021-06-10(Thu) 12:30~

  • seminar
    • Maruyama slide
  • tips
    • Liu research tips.pdf. [extra]
  • chair
    • Ma

2021-06-17(Thu) 12:30~

  • research
    • Yang Progress_report
    • Vijay Slides
  • tips
    • Sangwhan Slides
  • chair
    • Wiem

2021-06-24(Thu) 12:30~

  • research
    • Toyama Construction of a QA system on COVID-19 / NTCIR: Fact Verification

    • Niwa: Non-autoregressive Generation with Nearest Neighbors.pdf

  • tips
    • Wiem Networking and Opportunities Abroad
  • chair
    • Koyama

2021-07-01(Thu) 12:30~

  • research
    • Maeda : [Cracking of parameterised metrics]
    • Erick
  • tips
    • Koyama To write a paper without crisis
  • Chair
    • Ueki

2021-07-08(Thu) 12:30~

  • research
    • Ishikawa: slides
    • Hiraoka: My researches on Tokenization
  • tips
    • Ueki
  • chair
    • Maruyama

2021-07-15(Thu) 12:30~

  • research
    • Loem : pseudo-data for summarization task

    • Takase: Rethinking Perturbations in Encoder-Decoders for Fast Training

  • tips
    • Kaneko: Author response
  • chair
    • Kuo

2021-07-22(Thu) 12:30~

  • rehearsal of master-thesis presentation
    • Ma: Final_NER and RE using Contextualized Representations

    • Wiem Behavioral Testing of KGE models (Master Thesis)

  • tips
    • Yoshikawa Learning English from home
  • chair
    • Taniguchi

2021-07-29(Thu) 12:30~

  • research
    • Sangwhan: Not Really Research

    • Mizuki: Learning Hierarchical Code Representation for Hyponymy Detection

      • Introduction to the Evaluation of Lexical Semantic Relation
  • tips
    • Vijay: Productivity
      • How to Improve Productivity.pdf (22.0 MB)
  • chair
    • Toyama

2021-08-05(Thu) 12:30~

  • research
    • Iida Incorporating Semantic Textual Similarity and Lexical Matching for Information Retrieval

    • Kuo Learning Causality and Parts Relationship in Car Recall Text

  • tips
    • Erick: Docker & Terraform for reproducible research. Docker&Terraform PPT Docker&Terraform PDF
  • chair
    • Maeda

2021-09-02(Thu) 12:30~

  • reheasal
    • Ueki
      • mid-term-presentation
    • Koyama ニューラル文法誤り訂正のデータ拡張における多様な誤りの重要性

    • Kuo Learning Causality and Parts Relationship in Car Recall Text
  • chair
    • Ishikawa

2021-09-09(Thu) 12:30~

  • research
    • Niwa New Protocol to Create Benchmark Datasets for Societal Biases
    • Liu Semi-supervised Text Generation with Consistency Regularization
  • tips
    • Taniguchi How to launch services with GPU
  • chair
    • Loem

2021-09-22(Wed) 11:00~

  • research
    • Yang 2021_09_22.pdf (2.2 MB)

    • Muraoka: Fine-tuning OSCAR on IPP task

  • tips
    • Ma: My Strategies of Learning Japanese
  • chair
    • Koyama

2021-09-30(Thu) 12:30~

  • research
    • Takase: Strategies for Sharing Parameters across Layers

    • Taniguchi Personalization phrase generation

  • tips
    • Kuo: tools/sites I would like to share
  • chair
    • Maruyama

2021-10-07(Thu) 12:30~

  • research
    • Wiem
    • ErickVisual Language Navigation
  • tips
    • Toyama stress_coping
  • chair
    • Taniguchi

2021-10-14(Thu) 12:30~

  • research
    • Ma: label-dependency-incorporated joint entity and relation extraction
    • Ueki improve truthfulness of headline generation
  • tips
    • Maruyama
    • 2021-10-14_ResearchTips.pdf (818.2 kB)
  • chair
    • Kuo

2021-10-21(Thu) 12:30~

  • research
    • Hiraoka
    • Koyama: rule-augmented unsupervised pre-training for grammatical error correction
  • tips
    • Takase: How to progress research
  • chair
    • Toyama

2021-10-28(Thu) 12:30~

  • research
    • Iida Incorporating Semantic Textual Similarity and Lexical Matching for Information Retrieval

    • Mizuki: Learning Hierarchical Sense Code Representation for knowledge-based Word Sense Disambiguation

  • tips
    • Niwa: Make the best use of lab’s web page, esa page
  • chair
    • Maeda

2021-11-04(Thu) 12:30~

  • research[RS_2021_11_04.pdf (629.9 kB)]

    • Kaneko Grammatical Error Correction
    • Yoshikawa Evaluation of LAMA probe in selective prediction setting (2021.12.02 revised)
  • tips
    • Yang Install Kubuntu on a Windows machine
  • chair
    • Loem

2021-11-11(Thu) 12:30~

  • research
    • Maruyama
    • Vijay
      • Research Topics: Fluency and Template Summarization
  • tips
    • Hiraoka
      • Three Coding Tips for Research
  • chair
    • Ishikawa

2021-11-18(Thu) 12:30~

  • research
    • Kuo
    • Sangwhan
  • tips
    • Mizuki: Introduction to PyTorch Dataset class
  • chair
    • Ueki

2021-11-25(Thu) 12:30~

  • research
  • tips
    • Koyama data-bin in fairseq
  • chair
    • Maeda

2021-12-02(Thu) 12:30~

  • research
    • Maeda 学習者コーパスを用いた文法誤り訂正の自動評価

    • Ishikawa slides

  • tips
    • Iida data collater
  • chair
    • Koyama

2021-12-09(Thu) 12:30~

2021-12-16(Thu) 12:30~

  • research
    • Loem: Efficient Data Augmentation for Abstractive Summarization
    • Niwa: Non-autoregressive Generation with Nearest Neighbors
  • chair
    • Maruyama

2021-12-23(Thu) 12:30~

  • research
    • Liu Semi-supervised formality style transfer with consistency training.pdf (3.4 MB)

    • Yang Scientific Figure Captioning (1.2 MB)

  • tips
    • Sangwhan - slides
  • chair
    • Kuo

2022-01-20(Thu) 12:30~

  • research
    • Muraoka: Refactoring and Re-experiment (4.3 MB)

    • Taniguchi: Advertising phrase generation for a sense of ownership.pdf (584.9 kB)

  • chair
    • Loem

2022-01-27(Thu) 12:30~

  • research (master thesis presentation rehearsal) 20min
    • Ueki:slide
    • Koyama: slide
  • tips
    • Muraoka: My routine for reproducible experiments (687.5 kB)
  • chair
    • Ishikawa

2022-02-03(Thu) 12:30~

  • B4 thesis presentation rehearsal
    • Ishikawa 疑似訓練データによる格助詞の省略に頑健な係り受け解析

    • Maeda : パラレルデータにおける修正の影響度に基づいた文法誤り訂正の自動評価手法

    • Loem : 抽出型要約と言い換えによる生成型要約の訓練データ拡張

  • chair
    • Ueki

2022-03-03(Thu) 12:30~

Rehearsal Memo * 9~11月 * M2(春入学) 中間発表 (多分9月) * Ueki,Koyama * M2(秋入学) 修論発表 (多分9月) * Ma,Wiem * M1(春入学) 構想発表 (多分11月) * Miyazaki,Maruyama,Taniguchi * M1(秋入学) 構想発表(多分9月) * Kuo * 2~3月 * B4 卒論発表 * hoge * M2(春入学) 修論発表 * Ueki,Koyama * M1(秋入学) 中間発表 * Kuo * Doctor? * 言語処理学会