The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
Rangkaian saraf konvolusi berbilang tugas yang membawa kepada prestasi tinggi dan kebolehtafsiran melalui anggaran atribut dibentangkan dalam surat ini. Kaedah kami boleh memberikan tafsiran hasil klasifikasi CNN dengan mengeluarkan atribut yang menerangkan elemen objek sebagai alasan penghakiman CNN di lapisan tengah. Tambahan pula, rangkaian yang dicadangkan menggunakan atribut anggaran untuk ramalan kelas berikut. Akibatnya, pembinaan CNN pelbagai tugas baru dengan penambahbaikan dalam kedua-dua kebolehtafsiran dan prestasi klasifikasi direalisasikan.
Keisuke MAEDA
Hokkaido University
Kazaha HORII
Hokkaido University
Takahiro OGAWA
Hokkaido University
Miki HASEYAMA
Hokkaido University
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Salinan
Keisuke MAEDA, Kazaha HORII, Takahiro OGAWA, Miki HASEYAMA, "Multi-Task Convolutional Neural Network Leading to High Performance and Interpretability via Attribute Estimation" in IEICE TRANSACTIONS on Fundamentals,
vol. E103-A, no. 12, pp. 1609-1612, December 2020, doi: 10.1587/transfun.2020SML0006.
Abstract: 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.
URL: https://global.ieice.org/en_transactions/fundamentals/10.1587/transfun.2020SML0006/_p
Salinan
@ARTICLE{e103-a_12_1609,
author={Keisuke MAEDA, Kazaha HORII, Takahiro OGAWA, Miki HASEYAMA, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={Multi-Task Convolutional Neural Network Leading to High Performance and Interpretability via Attribute Estimation},
year={2020},
volume={E103-A},
number={12},
pages={1609-1612},
abstract={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.},
keywords={},
doi={10.1587/transfun.2020SML0006},
ISSN={1745-1337},
month={December},}
Salinan
TY - JOUR
TI - Multi-Task Convolutional Neural Network Leading to High Performance and Interpretability via Attribute Estimation
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 1609
EP - 1612
AU - Keisuke MAEDA
AU - Kazaha HORII
AU - Takahiro OGAWA
AU - Miki HASEYAMA
PY - 2020
DO - 10.1587/transfun.2020SML0006
JO - IEICE TRANSACTIONS on Fundamentals
SN - 1745-1337
VL - E103-A
IS - 12
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - December 2020
AB - 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.
ER -