• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to previous page

Machine Learning with Small Data

Research Project

Project/Area Number 19K22863
Research Category

Grant-in-Aid for Challenging Research (Exploratory)

Allocation TypeMulti-year Fund
Review Section Medium-sized Section 61:Human informatics and related fields
Research InstitutionThe University of Tokyo

Principal Investigator

Yamasaki Toshihiko  東京大学, 大学院情報理工学系研究科, 准教授 (70376599)

Project Period (FY) 2019-06-28 – 2022-03-31
Project Status Completed (Fiscal Year 2021)
Budget Amount *help
¥6,500,000 (Direct Cost: ¥5,000,000、Indirect Cost: ¥1,500,000)
Fiscal Year 2020: ¥3,250,000 (Direct Cost: ¥2,500,000、Indirect Cost: ¥750,000)
Fiscal Year 2019: ¥3,250,000 (Direct Cost: ¥2,500,000、Indirect Cost: ¥750,000)
Keywords少量学習データ / ドメイン適応 / 弱教師付き学習 / 半教師付き学習 / Few -Shot学習 / Zero-Shot学習 / 機械学習 / 転移学習 / 弱/半教師付き学習 / Few/Zero-Shot学習
Outline of Research at the Start

少量学習データに対応可能な学習アルゴリズムの実証と最適化・一般化を行っていく。具体的には、研究代表者が専門分野としている画像・映像を対象に下記の研究などを行っていく。(1) 転移学習と弱教師あり学習の組み合わせによるドメイン適応、(2) 学習曲線に注目した弱教師あり学習、(3) Few-Shot学習におけるData Augmentation、(4) マルチモーダルを活用したZero-Shot学習、(5) 上記課題の一般化およびオープンソース化。

Outline of Final Research Achievements

Deep learning is known for its overwhelming performance, but such high performance can be achieved only when large amounts of correct data are available. On the other hand, the time and financial cost for acquiring and creating such data is a severe problem. While rich data sets are available in fields where research has already been conducted for a long time, such as object recognition, large-scale datasets cannot be expected for industries or medical fields. In order to solve this problem, we have achieved various elemental technologies such as domain adaptation, weak/semi-supervised learning, few/zero-shot learning, self-supervised learning, and unbalanced learning that enable us to obtain robust machine learning models even with small and biased datasets.

Academic Significance and Societal Importance of the Research Achievements

前述の通り、深層学習アルゴリズムがその性能を十分に発揮するためには、大規模かつ性格なラベリングがなされた学習データが必要であった。産業応用を考えた場合、新たに大量のデータを収集し、かつそのデータに正解ラベルを付与するコストはあまりにも高い。本研究成果はこの制約を緩和し、少量であったり極端にデータの分布に偏りがあったりしても正しく学習がなされる種々の要素技術を実現したことにある。我々の技術はすでにいくつかは実サービスに応用されており、社会的にも貢献している。

Report

(4 results)
  • 2021 Annual Research Report   Final Research Report ( PDF )
  • 2020 Research-status Report
  • 2019 Research-status Report
  • Research Products

    (36 results)

All 2022 2021 2020 2019 Other

All Int'l Joint Research (2 results) Journal Article (9 results) (of which Peer Reviewed: 9 results,  Open Access: 6 results) Presentation (23 results) (of which Int'l Joint Research: 15 results,  Invited: 1 results) Remarks (1 results) Patent(Industrial Property Rights) (1 results)

  • [Int'l Joint Research] 清華大学(中国)

    • Related Report
      2020 Research-status Report
  • [Int'l Joint Research] 清華大学(中国)

    • Related Report
      2019 Research-status Report
  • [Journal Article] Assessment System of Presentation Slide Design Using Visual and Structural Features2022

    • Author(s)
      Shengzhou Yi, Junichiro Matsugami, and Toshihiko Yamasaki
    • Journal Title

      IEICE Transactions on Information and Systems

      Volume: E105.D Issue: 3 Pages: 587-596

    • DOI

      10.1587/transinf.2021HCK0001

    • NAID

      130008165598

    • ISSN
      0916-8532, 1745-1361
    • Year and Date
      2022-03-01
    • Related Report
      2021 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] An Improved Inter-intra Contrastive Learning Framework on Self-supervised Video Representation2022

    • Author(s)
      Li Tao, Xueting Wang, and Toshihiko Yamasaki
    • Journal Title

      IEEE Transactions on Circuits and Systems for Video Technology

      Volume: NA Issue: 8 Pages: 5266-5280

    • DOI

      10.1109/tcsvt.2022.3141051

    • Related Report
      2021 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Robustness of Deep Learning Models in Dermatological Evaluation: A Critical Assessment2021

    • Author(s)
      Sourav Mishra, Subhajit Chaudhary, Hideaki Imaizumi, and Toshihiko Yamasaki
    • Journal Title

      IEICE Transactions on Information and Systems

      Volume: E104.D Issue: 3 Pages: 419-429

    • DOI

      10.1587/transinf.2020EDP7133

    • NAID

      130007993191

    • ISSN
      0916-8532, 1745-1361
    • Year and Date
      2021-03-01
    • Related Report
      2020 Research-status Report
    • Peer Reviewed / Open Access
  • [Journal Article] Rethinking Motion Representation: Residual Frames with 3D ConvNets2021

    • Author(s)
      Li Tao, Xueting Wang, and Toshihiko Yamasaki
    • Journal Title

      IEEE Transactions on Image Processing

      Volume: 30 Pages: 9231-9244

    • DOI

      10.1109/tip.2021.3124156

    • Related Report
      2021 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Object-aware Instance Labeling for Weakly Supervised Object Detection2020

    • Author(s)
      Satoshi Kosugi and Toshihiko Yamasaki
    • Journal Title

      NIHON GAZO GAKKAISHI (Journal of the Imaging Society of Japan)

      Volume: 59 Issue: 6 Pages: 585-590

    • DOI

      10.11370/isj.59.585

    • NAID

      130007952865

    • ISSN
      1344-4425, 1880-4675
    • Year and Date
      2020-12-10
    • Related Report
      2020 Research-status Report
    • Peer Reviewed
  • [Journal Article] Toward Better Planetary Surface Exploration by Mars Orbital Imagery Inpainting2020

    • Author(s)
      Hiya Roy, Subhajit Chaudhury, Toshihiko Yamasaki, and Tatsuaki Hashimoto
    • Journal Title

      IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

      Volume: - Pages: 175-189

    • DOI

      10.1109/jstars.2020.3038778

    • Related Report
      2020 Research-status Report
    • Peer Reviewed
  • [Journal Article] Learning from Synthetic Shadows for Shadow Detection and Removal2020

    • Author(s)
      Naoto Inoue and Toshihiko Yamasaki
    • Journal Title

      IEEE Transactions on Circuits and Systems for Video Technology

      Volume: - Issue: 11 Pages: 4187-4197

    • DOI

      10.1109/tcsvt.2020.3047977

    • Related Report
      2020 Research-status Report
    • Peer Reviewed / Open Access
  • [Journal Article] Benchmarking robustness in deep learning under adversarial distribution shifts in medical image analysis2020

    • Author(s)
      Subhajit Chaudhury and Toshihiko Yamasaki
    • Journal Title

      IEEE Access

      Volume: 9 Pages: 37039-37053

    • DOI

      10.1109/access.2021.3062990

    • Related Report
      2020 Research-status Report
    • Peer Reviewed / Open Access
  • [Journal Article] Image Inpainting using Frequency Domain Priors2020

    • Author(s)
      Hiya Roy, Subhajit Chaudhury, Toshihiko Yamasaki, and Tatsuaki Hashimoto
    • Journal Title

      SPIE Journal of Electronic Imaging

      Volume: 30 Issue: 02

    • DOI

      10.1117/1.jei.30.2.023016

    • Related Report
      2020 Research-status Report
    • Peer Reviewed
  • [Presentation] Reproducibility Companion Paper: Self-supervised Video Representation Learning Using Inter-intra Contrastive Framework2021

    • Author(s)
      Li Tao, Xueting Wang, Toshihiko Yamasaki, Jingjing Chen, and Steven Hicks
    • Organizer
      The 29th ACM International Conference on Multimedia
    • Related Report
      2021 Annual Research Report
    • Int'l Joint Research
  • [Presentation] An Improved Inter-Intra Contrastive Framework for Self-Supervised Video Representation Learning2021

    • Author(s)
      Li Tao, Xueting Wang, and Toshihiko Yamasaki
    • Organizer
      第24回画像の認識・理解シンポジウム
    • Related Report
      2021 Annual Research Report
  • [Presentation] Better Temporal Representation for Unsupervised Video Summarization Based on Contrastive Self-Supervised Learning2021

    • Author(s)
      Xianliang Zhang, Li Tao, Xueting Wang, and Toshihiko Yamasaki
    • Organizer
      第24回画像の認識・理解シンポジウム
    • Related Report
      2021 Annual Research Report
  • [Presentation] Class-Balanced Contrastive Pre-Training for Improving Long-Tailed Recognition2021

    • Author(s)
      Shengzhou Yi, Li Tao, Xueting Wang, and Toshihiko Yamasaki
    • Organizer
      第24回画像の認識・理解シンポジウム
    • Related Report
      2021 Annual Research Report
  • [Presentation] Identifying Design Problems of Presentation Slides using a Bimodal Neural Network2021

    • Author(s)
      Shengzhou Yi, Junichiro Matsugami, and Toshihiko Yamasaki
    • Organizer
      メディアエクスペリエンス・バーチャル環境基礎研究会
    • Related Report
      2021 Annual Research Report
  • [Presentation] Pretext-Contrastive Learning for Self-Supervised Video Representation Learning2021

    • Author(s)
      Li Tao, Xueting Wang, and Toshihiko Yamasaki
    • Organizer
      信学技報, 画像工学研究会
    • Related Report
      2021 Annual Research Report
  • [Presentation] Contrastive Self-Supervised Learning Framework for Unsupervised Video Summarization2021

    • Author(s)
      Xianliang Zhang, Li Tao, Xueting Wang, and Toshihiko Yamasaki
    • Organizer
      信学技報, 画像工学研究会
    • Related Report
      2021 Annual Research Report
  • [Presentation] 行動検出タスクにおける人物特徴量に注目した自己教師あり学習2021

    • Author(s)
      吉田舜, 山崎俊彦
    • Organizer
      情報処理学会第84 回全国大会
    • Related Report
      2021 Annual Research Report
  • [Presentation] Investigating generalization in neural networks under optimally evolved training perturbations2020

    • Author(s)
      Subhajit Chaudhury and Toshihiko Yamasaki
    • Organizer
      The 45th International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2020)
    • Related Report
      2020 Research-status Report
    • Int'l Joint Research
  • [Presentation] Weakly supervised segmentation guided hand pose estimation during interaction with unknown objects2020

    • Author(s)
      Cairong Zhang, Guijin Wang, Xinghao Chen, Pengwei Xie, and Toshihiko Yamasaki
    • Organizer
      The 45th International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2020)
    • Related Report
      2020 Research-status Report
    • Int'l Joint Research
  • [Presentation] Assessing Robustness of Deep Learning Methods in Dermatological Workflow2020

    • Author(s)
      Sourav Mishra, Subhajit Chaudhary, Hideaki Imaizumi, and Toshihiko Yamasaki
    • Organizer
      Workshop of the ACM Conference on Health, Inference, and Learning (CHIL)
    • Related Report
      2020 Research-status Report 2019 Research-status Report
    • Int'l Joint Research
  • [Presentation] Self-supervised Video Representation Learning Using Inter-intra Contrastive Framework2020

    • Author(s)
      Li Tao, Xueting Wang, and Toshihiko Yamasaki
    • Organizer
      the 28th ACM International Conference on Multimedia (MM '20)
    • Related Report
      2020 Research-status Report
    • Int'l Joint Research
  • [Presentation] Motion Representation Using Residual Frames with 3D CNN2020

    • Author(s)
      Li Tao, Xueting Wang and T. Yamasaki
    • Organizer
      2020 IEEE International Conference on Image Processing (ICIP)
    • Related Report
      2020 Research-status Report
    • Int'l Joint Research
  • [Presentation] Feature Point Matching in Cross-Spectral Images with Cycle Consistency Learning2020

    • Author(s)
      Ryosuke Furuta, Naoaki Noguchi, Xueting Wang, Toshihiko Yamasaki
    • Organizer
      2020 25th International Conference on Pattern Recognition (ICPR)
    • Related Report
      2020 Research-status Report
    • Int'l Joint Research
  • [Presentation] Very Important Person Localization in Unconstrained Conditions: A New Benchmark2020

    • Author(s)
      Xiao Wang, Zheng Wang, Toshihiko Yamasaki, and Wenjun Zeng
    • Organizer
      The Thirty-Fifth AAAI Conference on Artificial Intelligence (AAAI-21)
    • Related Report
      2020 Research-status Report
    • Int'l Joint Research
  • [Presentation] Improving image classifiers for small datasets by learning rate adaptations2019

    • Author(s)
      Sourav Mishra, Toshihiko Yamasaki, and Hideaki Imaizumi
    • Organizer
      The 16th International Conference on Machine Vision Applications (MVA)
    • Related Report
      2019 Research-status Report
    • Int'l Joint Research
  • [Presentation] Interpreting fine-grained dermatological classification by Deep learning2019

    • Author(s)
      Sourav Mishra, Hideaki Imaizumi, and Toshihiko Yamasaki
    • Organizer
      ISIC Skin Image Analysis Workshop at CVPR
    • Related Report
      2019 Research-status Report
    • Int'l Joint Research
  • [Presentation] Reinforcing the robustness of a deep neural network to adversarial examples by using color quantization of training image data2019

    • Author(s)
      Shuntaro Miyazato, Xueting WANG, Toshihiko Yamasaki, and Kiyoharu Aizawa
    • Organizer
      2019 IEEE International Conference on Image (ICIP)
    • Related Report
      2019 Research-status Report
    • Int'l Joint Research
  • [Presentation] Zero-Shot Semantic Segmentation via Variational Mapping2019

    • Author(s)
      Naoki Kato, Toshihiko Yamasaki, and Kiyoharu Aizawa
    • Organizer
      ICCV 2019 Workshop on Multi-Discipline Approach for Learning Concepts - Zero-Shot One-Shot Few-Shot and Beyond
    • Related Report
      2019 Research-status Report
    • Int'l Joint Research
  • [Presentation] Object-aware instance labeling for weakly supervised object detection2019

    • Author(s)
      Satoshi Kosugi, Toshihiko Yamasaki, and Kiyoharu Aizawa
    • Organizer
      International Conference on Computer Vision (ICCV)
    • Related Report
      2019 Research-status Report
    • Int'l Joint Research
  • [Presentation] Weakly Supervised Video Summarization by Hierarchical Reinforcement Learning2019

    • Author(s)
      Yiyan Chen, Li Tao, Xueting Wang, and Toshihiko Yamasaki
    • Organizer
      ACMMM Asia
    • Related Report
      2019 Research-status Report
    • Int'l Joint Research
  • [Presentation] Unpaired Image Enhancement Featuring Reinforcement-Learning-Controlled Image Editing Software2019

    • Author(s)
      Satoshi Kosugi and Toshihiko Yamasaki
    • Organizer
      The Thirty-Fourth AAAI Conference on Artificial Intelligence (AAAI)
    • Related Report
      2019 Research-status Report
    • Int'l Joint Research
  • [Presentation] 情報科学者から見た医療AIの現状と可能性2019

    • Author(s)
      山崎俊彦
    • Organizer
      第59回日本核医学会総会
    • Related Report
      2019 Research-status Report
    • Invited
  • [Remarks]

    • URL

      https://www.hal.t.u-tokyo.ac.jp/~yamasaki/project/index-e.html

    • Related Report
      2019 Research-status Report
  • [Patent(Industrial Property Rights)] 画像解析装置、画像解析方法及び画像解析プログラム2019

    • Inventor(s)
      山崎俊彦, 小杉哲
    • Industrial Property Rights Holder
      山崎俊彦, 小杉哲
    • Industrial Property Rights Type
      特許
    • Industrial Property Number
      2019-082100
    • Filing Date
      2019
    • Related Report
      2019 Research-status Report

URL: 

Published: 2019-07-04   Modified: 2023-01-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi