TY - GEN
T1 - More is Not Always Better
T2 - 2024 IEEE/ACM International Workshop on Deep Learning for Testing and Testing for Deep Learning, DeepTest 2024
AU - Mancu, Andrei
AU - Laurent, Thomas
AU - Rieger, Franz
AU - Arcaini, Paolo
AU - Ishikawa, Fuyuki
AU - Rueckert, Daniel
N1 - Publisher Copyright:
© 2024 Copyright is held by the owner/author(s). Publication rights licensed to ACM.
PY - 2024/4/20
Y1 - 2024/4/20
N2 - DNN repair is an effective technique applied after training to enhance the class-specific accuracy of classifier models, where a low failure rate is required on specific classes. The repair methods introduced in recent studies assume that they are applied to fully trained models. In this paper, we argue that this could not always be the best choice. We analyse the performance of DNN models under various training times and repair combinations. Through meticulously designed experiments on two real-world datasets and a carefully curated assessment score, we show that applying DNN repair earlier in the training process, and not only at its end, can be beneficial. Thus, we encourage the research community to consider the idea of when to apply DNN repair in the model development.
AB - DNN repair is an effective technique applied after training to enhance the class-specific accuracy of classifier models, where a low failure rate is required on specific classes. The repair methods introduced in recent studies assume that they are applied to fully trained models. In this paper, we argue that this could not always be the best choice. We analyse the performance of DNN models under various training times and repair combinations. Through meticulously designed experiments on two real-world datasets and a carefully curated assessment score, we show that applying DNN repair earlier in the training process, and not only at its end, can be beneficial. Thus, we encourage the research community to consider the idea of when to apply DNN repair in the model development.
KW - DNN repair
KW - DNN training
KW - deep neural networks
KW - safety-critical
UR - http://www.scopus.com/inward/record.url?scp=85194856096&partnerID=8YFLogxK
U2 - 10.1145/3643786.3648024
DO - 10.1145/3643786.3648024
M3 - Conference contribution
AN - SCOPUS:85194856096
T3 - Proceedings - 2024 IEEE/ACM International Workshop on Deep Learning for Testing and Testing for Deep Learning, DeepTest 2024
SP - 13
EP - 16
BT - Proceedings - 2024 IEEE/ACM International Workshop on Deep Learning for Testing and Testing for Deep Learning, DeepTest 2024
PB - Association for Computing Machinery, Inc
Y2 - 20 April 2024
ER -