Multi-Task Convolutional Neural Network Leading to High Performance and Interpretability via Attribute Estimation

• MAEDA Keisuke

  Office of Institutional Research, Hokkaido University

• HORII Kazaha

  Graduate School of Information Science and Technology, Hokkaido University

• OGAWA Takahiro

  Faculty of Information Science and Technology, Hokkaido University

• HASEYAMA Miki

  Faculty of Information Science and Technology, Hokkaido University

抄録

A multi-task convolutional neural network leading to high performance and interpretability via attribute estimation is presented in this letter. Our method can provide interpretation of the classification results of CNNs by outputting attributes that explain elements of objects as a judgement reason of CNNs in the middle layer. Furthermore, the proposed network uses the estimated attributes for the following prediction of classes. Consequently, construction of a novel multi-task CNN with improvements in both of the interpretability and classification performance is realized.

収録刊行物

• IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences

  IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences E103.A (12), 1609-1612, 2020-12-01

  一般社団法人 電子情報通信学会

キーワード

• multi-task convolutional neural network
• image classification
• attribute estimation
• interpretability

詳細情報

• CRID: 1390286426511814912
• NII論文ID: 130007948361
• DOI: 10.1587/transfun.2020sml0006
• ISSN: 17451337
• ISSN: 09168508
• HANDLE: 2115/80378
• Web Site: https://www.jstage.jst.go.jp/article/transfun/E103.A/12/E103.A_2020SML0006/_pdf
• 本文言語コード: en
• データソース種別:
  • JaLC
  • IRDB
  • Crossref
  • CiNii Articles
  • KAKEN
• 抄録ライセンスフラグ: 使用不可