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
본 논문에서는 시계열 측정 데이터로부터 변속 품질을 평가하기 위해 심층 신경망을 사용하는 CSQ-SDL이라는 최근 제안된 방법에 기계 학습의 두 가지 방법인 드롭아웃과 준지도 학습을 적용합니다. 새로운 자동변속기(AT)를 개발할 때, 구경 측정 전 세계 모든 도로에서 발생하는 모든 상황에서 쾌적한 운전 경험을 실현하기 위해 AT의 많은 매개변수를 조정하는 과정이 진행됩니다. 교정을 위해서는 매개변수가 변경될 때마다 전문가가 실험의 시계열 측정 데이터에서 이동 품질을 시각적으로 평가해야 하며 이는 반복적이고 시간이 많이 소요됩니다. CSQ-SDL은 시각적 평가에 소요되는 시간을 단축하기 위해 개발되었으며, 그 효과는 충분한 수의 데이터 포인트를 획득하는 데 달려 있습니다. 그러나 실제로는 데이터 양이 부족한 경우가 많습니다. 여기서 제안하는 방법은 이러한 경우를 처리할 수 있습니다. 라벨링된 데이터 포인트 수가 적은 경우에는 드롭아웃을 사용하는 방법을 제안합니다. 레이블이 있는 데이터 포인트의 개수는 적지만 레이블이 지정되지 않은 데이터의 개수가 충분한 경우에는 준지도 학습(semi-supervised learning)을 사용하는 방법을 제안합니다. 실험에 따르면 전자는 중간 수준의 성능 향상을 제공하는 반면 후자는 상당한 성능 향상을 제공하는 것으로 나타났습니다.
Takefumi KAWAKAMI
AISIN CORPORATION
Takanori IDE
AISIN CORPORATION
Kunihito HOKI
The University of Electro-Communications
Masakazu MURAMATSU
The University of Electro-Communications
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.
부
Takefumi KAWAKAMI, Takanori IDE, Kunihito HOKI, Masakazu MURAMATSU, "Shift Quality Classifier Using Deep Neural Networks on Small Data with Dropout and Semi-Supervised Learning" in IEICE TRANSACTIONS on Information,
vol. E106-D, no. 12, pp. 2078-2084, December 2023, doi: 10.1587/transinf.2023EDP7033.
Abstract: In this paper, we apply two methods in machine learning, dropout and semi-supervised learning, to a recently proposed method called CSQ-SDL which uses deep neural networks for evaluating shift quality from time-series measurement data. When developing a new Automatic Transmission (AT), calibration takes place where many parameters of the AT are adjusted to realize pleasant driving experience in all situations that occur on all roads around the world. Calibration requires an expert to visually assess the shift quality from the time-series measurement data of the experiments each time the parameters are changed, which is iterative and time-consuming. The CSQ-SDL was developed to shorten time consumed by the visual assessment, and its effectiveness depends on acquiring a sufficient number of data points. In practice, however, data amounts are often insufficient. The methods proposed here can handle such cases. For the cases wherein only a small number of labeled data points is available, we propose a method that uses dropout. For those cases wherein the number of labeled data points is small but the number of unlabeled data is sufficient, we propose a method that uses semi-supervised learning. Experiments show that while the former gives moderate improvement, the latter offers a significant performance improvement.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2023EDP7033/_p
부
@ARTICLE{e106-d_12_2078,
author={Takefumi KAWAKAMI, Takanori IDE, Kunihito HOKI, Masakazu MURAMATSU, },
journal={IEICE TRANSACTIONS on Information},
title={Shift Quality Classifier Using Deep Neural Networks on Small Data with Dropout and Semi-Supervised Learning},
year={2023},
volume={E106-D},
number={12},
pages={2078-2084},
abstract={In this paper, we apply two methods in machine learning, dropout and semi-supervised learning, to a recently proposed method called CSQ-SDL which uses deep neural networks for evaluating shift quality from time-series measurement data. When developing a new Automatic Transmission (AT), calibration takes place where many parameters of the AT are adjusted to realize pleasant driving experience in all situations that occur on all roads around the world. Calibration requires an expert to visually assess the shift quality from the time-series measurement data of the experiments each time the parameters are changed, which is iterative and time-consuming. The CSQ-SDL was developed to shorten time consumed by the visual assessment, and its effectiveness depends on acquiring a sufficient number of data points. In practice, however, data amounts are often insufficient. The methods proposed here can handle such cases. For the cases wherein only a small number of labeled data points is available, we propose a method that uses dropout. For those cases wherein the number of labeled data points is small but the number of unlabeled data is sufficient, we propose a method that uses semi-supervised learning. Experiments show that while the former gives moderate improvement, the latter offers a significant performance improvement.},
keywords={},
doi={10.1587/transinf.2023EDP7033},
ISSN={1745-1361},
month={December},}
부
TY - JOUR
TI - Shift Quality Classifier Using Deep Neural Networks on Small Data with Dropout and Semi-Supervised Learning
T2 - IEICE TRANSACTIONS on Information
SP - 2078
EP - 2084
AU - Takefumi KAWAKAMI
AU - Takanori IDE
AU - Kunihito HOKI
AU - Masakazu MURAMATSU
PY - 2023
DO - 10.1587/transinf.2023EDP7033
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E106-D
IS - 12
JA - IEICE TRANSACTIONS on Information
Y1 - December 2023
AB - In this paper, we apply two methods in machine learning, dropout and semi-supervised learning, to a recently proposed method called CSQ-SDL which uses deep neural networks for evaluating shift quality from time-series measurement data. When developing a new Automatic Transmission (AT), calibration takes place where many parameters of the AT are adjusted to realize pleasant driving experience in all situations that occur on all roads around the world. Calibration requires an expert to visually assess the shift quality from the time-series measurement data of the experiments each time the parameters are changed, which is iterative and time-consuming. The CSQ-SDL was developed to shorten time consumed by the visual assessment, and its effectiveness depends on acquiring a sufficient number of data points. In practice, however, data amounts are often insufficient. The methods proposed here can handle such cases. For the cases wherein only a small number of labeled data points is available, we propose a method that uses dropout. For those cases wherein the number of labeled data points is small but the number of unlabeled data is sufficient, we propose a method that uses semi-supervised learning. Experiments show that while the former gives moderate improvement, the latter offers a significant performance improvement.
ER -