presentations

Talks and Posters (Physical/Virtual)

統計的因果モデルに基づく転移学習 (invited)

Venue: 第17回日本統計学会春季集会
Date: March 05, 2023
Location: 1F Room 120 Zoom A
Authors: Takeshi Teshima
Presenter: Takeshi Teshima
Links:
- slides (Japanese)

データ拡張による因果グラフ的事前知識の予測モデリングへの活用

Venue: 第24回情報論的学習理論ワークショップ
Date: November 13, 2021
Location: Online (Zoom)
Authors: Takeshi Teshima and Masashi Sugiyama
Presenter: Takeshi Teshima
Links:
- slides (Japanese)

可逆ニューラルネットのSobolev空間における普遍性について

Venue: 第24回情報論的学習理論ワークショップ
Date: November 11, 2021
Location: Online (Zoom)
Authors: 石川勲, 手嶋毅志, 東條広一, 大野健太, 池田正弘, 杉山将
Presenter: 石川勲

γ-ABC: 頑健なダイバージェンス推定量を用いた外れ値ロバストな近似ベイズ計算

Venue: 第24回情報論的学習理論ワークショップ
Date: November 10, 2021
Location: Online (Zoom)
Authors: 藤澤将広, 手嶋毅志, 佐藤一誠, 杉山将
Presenter: 藤澤将広

Incorporating causal graphical prior knowledge into predictive modeling via simple data augmentation

Type: Virtual poster presentation
Venue: 37th Conference on Uncertainty in Artificial Intelligence
Date: July 28, 2021
Location: Online
Authors: Takeshi Teshima, Masashi Sugiyama
Presenter: Takeshi Teshima
Other Links:
- session (51 II-A3)
- video (Underline)

Lessons Learned from Big Data Project in Bangsamoro

Venue: KIC Webinar Series: ICT x Social Investment in The Philippines
Date: July 15, 2021
Location: Online (Zoom)
Authors: Ayaka Oishi, Takeshi Teshima
Presenter: Ayaka Oishi, Takeshi Teshima
Links:
- video (Japanese)

因果的知識は予測モデルの役に立つか？ (invited)

Venue: DL輪読会-外部セミナー
Date: July 02, 2021
Location: Online (Zoom)
Authors: Takeshi Teshima
Presenter: Takeshi Teshima
Links:
- slides (Japanese)

因果メカニズム転移による小標本ドメイン適応 (invited)

Venue: 情報論的学習理論と機械学習研究会（IBISML）オーガナイズドセッション『異なるタスクを活用する機械学習：転移学習，メタ学習』
Date: March 05, 2021
Location: Online (Zoom)
Authors: Takeshi Teshima
Presenter: Takeshi Teshima
Links:
- slides (Japanese)

On the Universality of Invertible Neural Networks (invited)

Venue: Workshop on Functional Inference and Machine Intelligence
Date: March 03, 2021
Location: Online (Zoom)
Authors: Takeshi Teshima, and Isao Ishikawa
Presenter: Takeshi Teshima, Isao Ishikawa
Links:
- Slides

Universal Approximation Property of Neural Ordinary Differential Equations

Type: Poster presentation
Venue: NeurIPS2020 Workshop: Differential Geometry meets Deep Learning (DiffGeo4DL)
Date: December 11, 2020
Location: Online
Authors: Takeshi Teshima, Koichi Tojo, Masahiro Ikeda, Isao Ishikawa, and Kenta Oono.
Presenter: Takeshi Teshima
Links:
- slides
- poster

Coupling-based Invertible Neural Networks Are Universal Diffeomorphism Approximators

Type: Poster presentation
Venue: Thirty-fourth Conference on Neural Information Processing Systems
Date: December 09, 2020
Location: Online
Authors: Takeshi Teshima, Isao Ishikawa, Koichi Tojo, Kenta Oono, Masahiro Ikeda, and Masashi Sugiyama
Presenter: Takeshi Teshima
Links:
- link
- poster

Coupling-based invertible neural networks are universal diffeomorphism approximators

Type: Oral presentation
Venue: Thirty-fourth Conference on Neural Information Processing Systems
Date: December 09, 2020
Location: Online
Authors: Takeshi Teshima, Isao Ishikawa, Koichi Tojo, Kenta Oono, Masahiro Ikeda, and Masashi Sugiyama
Presenter: Takeshi Teshima
Links:
- slides

Forecasting Internally Displaced Persons’ Movements with Artificial Intelligence

Venue: JASID 2020 (国際開発学会第31回大会; Domestic conference)
Date: December 06, 2020
Location: Online (Zoom)

カップリングに基づく可逆神経回路網は微分同相写像の万能近似器である

Type: Poster presentation
Venue: 第23回情報論的学習理論ワークショップ (IBIS2020)
Date: November 26, 2020
Location: Online
Authors: Takeshi Teshima, Isao Ishikawa, Koichi Tojo, Kenta Oono, Masahiro Ikeda, and Masashi Sugiyama
Presenter: Takeshi Teshima
Links:
- poster (Japanese)
Other Links:
- video (Japanese)

Coupling flow-based Invertible Neural Networks Are Diffeomorphism Approximators

Venue: Mathiine
Date: July 22, 2020
Location: Online (Zoom)
Authors: Takeshi Teshima
Presenter: Takeshi Teshima
Links:
- slides

Few-shot Domain Adaptation by Causal Mechanism Transfer (invited)

Venue: CFML勉強会
Date: July 22, 2020
Location: Online (Zoom)
Authors: Takeshi Teshima
Presenter: Takeshi Teshima
Links:
- slides (Japanese)

Few-shot Domain Adaptation by Causal Mechanism Transfer

Type: Virtual poster presentation
Venue: 37th International Conference on Machine Learning
Date: July 17, 2020
Location: Online
Authors: Takeshi Teshima, Issei Sato, Masashi Sugiyama
Presenter: Takeshi Teshima
Links:
- slides

Few-shot Domain Adaptation by Causal Mechanism Transfer

Type: Virtual poster presentation
Venue: Machine Learning Summer School (MLSS) 2020
Date: July 03, 2020
Location: Online
Authors: Takeshi Teshima
Presenter: Takeshi Teshima
Other Links:
- video

構造的因果モデルの類似性に基づくドメイン適応

Venue: Mathiine-learning
Date: March 29, 2020
Location: Zoom (online)
Presenter: Takeshi Teshima

構造的因果モデルの類似性に基づくドメイン適応 (invited)

Venue: 第三回理研AIP数学系合同セミナー
Date: February 20, 2020
Location: Tokyo, Japan
Authors: Takeshi Teshima
Presenter: Takeshi Teshima

打ち切り行列の補完：天井効果の治療法 (invited)

Venue: 第4回統計・機械学習若手シンポジウム
Date: November 15, 2019
Location: Nagoya, Japan
Authors: Takeshi Teshima
Presenter: Takeshi Teshima
Links:
- slides (Japanese)

Clipped matrix completion: a remedy for ceiling effects

Type: Poster presentation
Venue: Joint Workshop of BBDC, BZML and RIKEN AIP
Date: September 10, 2019
Location: Berlin, Germany
Authors: Takeshi Teshima
Presenter: Takeshi Teshima
Links:
- poster

上限つきRAE画像のベイズ推定

Venue: 統計数理研究所データサイエンス共同利用基盤施設研究集会「機械学習とデータ同化による木星磁気圏のダイナミクスの理解」
Date: July 27, 2019
Location: 東北大学理学部　合同C棟２F　N204（多目的室）
Authors: Takeshi Teshima
Presenter: Takeshi Teshima

Clipped Matrix Completion: A Remedy for Ceiling Effects

Venue: IRCN Math/Computation Seminar
Date: July 10, 2019
Location: Tokyo, Japan
Presenter: Takeshi Teshima

Clipped Matrix Completion

Venue: 日本IBM基礎研究所
Date: April 25, 2019
Location: Tokyo, Japan
Authors: Takeshi Teshima
Presenter: Takeshi Teshima

Clipped matrix completion: a remedy for ceiling effects

Type: Poster presentation
Venue: The Second Korea-Japan Machine Learning Workshop
Organizer: Yung-Kyun Noh (SNU), Helen Hyunjung Shin (Ajou U.), Masashi Sugiyama (RIKEN-AIP / The University of Tokyo)
Date: February 23, 2019
Location: Jeju, Korea
Authors: Takeshi Teshima
Presenter: Takeshi Teshima
Links:
- poster

研究分野の紹介・キャリア講座

Venue: 開智日本橋中学・高等学校
Date: February 08, 2019
Location: Tokyo, Japan
Authors: Takeshi Teshima
Presenter: Takeshi Teshima

Clipped matrix completion: a remedy for ceiling effects

Type: Poster presentation
Venue: Thirty-Third AAAI Conference on Artificial Intelligence
Organizer: The Association for the Advancement of Artificial Intelligence
Date: January 31, 2019
Location: Honolulu, Hawaii
Authors: Takeshi Teshima, Miao Xu, Issei Sato, Masashi Sugiyama
Presenter: Takeshi Teshima
Links:
- poster
- spotlight

打ち切り行列補完：天井効果への処方箋 (in Japanese)

Venue: Toshiba R&D Center
Date: January 11, 2019
Location: Kawasaki, Japan
Authors: Takeshi Teshima
Presenter: Takeshi Teshima

Clipped matrix completion (in Japanese)

Venue: Math-iine Learning
Date: December 15, 2018
Location: Tokyo, Japan
Authors: Takeshi Teshima
Presenter: Takeshi Teshima

打ち切り行列の補完：天井効果への処方箋

Type: Poster presentation
Venue: 第21回情報論的学習理論ワークショップ (IBIS 2018)
Organizer: 情報論的学習理論と機械学習研究会 (IBISML)
Date: November 06, 2018
Location: Hokkaido, Japan
Authors: Takeshi Teshima, Miao Xu, Issei Sato, and Masashi Sugiyama
Presenter: Takeshi Teshima
Links:
- poster (Japanese)

Clipped matrix completion

Type: Poster presentation
Venue: International Research Center for Neurointelligence (IRCN) Site Visit Poster Session
Date: June 19, 2018
Location: Tokyo, Japan
Authors: Takeshi Teshima
Presenter: Takeshi Teshima