TY - GEN
T1 - Tackling Model Mismatch with Mixup Regulated Test-Time Training
AU - Zhang, Bochao
AU - Shao, Rui
AU - Du, Jingda
AU - Yuen, Pc
AU - Luo, Wei
N1 - Publisher Copyright:
© 2023 IEEE.
PY - 2023/10/9
Y1 - 2023/10/9
N2 - Test-time training (TTT) is an emerging approach for addressing the problem of domain shift. In its framework, a test-time training phase is inserted between the training phase and the test phase. During the test-time training phase, the representation layers are adapted using an auxiliary task. Then the updated model will be used in the test phase. Although the idea is very intuitive, TTT does not demonstrate competitive performance compared with some other domain adaption methods. In this paper, we present both theoretical and empirical analyses to explain the subpar performance of TTT. In particular, we point out that TTT causes a new kind of problem, which we term as Model Mismatch. To address this problem of Model Mismatch, we analyse a simple yet effective method inspired by the idea of mixup in robust training. Such effectiveness is shown in the experimental results.
AB - Test-time training (TTT) is an emerging approach for addressing the problem of domain shift. In its framework, a test-time training phase is inserted between the training phase and the test phase. During the test-time training phase, the representation layers are adapted using an auxiliary task. Then the updated model will be used in the test phase. Although the idea is very intuitive, TTT does not demonstrate competitive performance compared with some other domain adaption methods. In this paper, we present both theoretical and empirical analyses to explain the subpar performance of TTT. In particular, we point out that TTT causes a new kind of problem, which we term as Model Mismatch. To address this problem of Model Mismatch, we analyse a simple yet effective method inspired by the idea of mixup in robust training. Such effectiveness is shown in the experimental results.
KW - mixup
KW - model mismatch
KW - test-time training
UR - http://www.scopus.com/inward/record.url?scp=85179010371&partnerID=8YFLogxK
U2 - 10.1109/DSAA60987.2023.10302565
DO - 10.1109/DSAA60987.2023.10302565
M3 - Conference proceeding
AN - SCOPUS:85179010371
T3 - 2023 IEEE 10th International Conference on Data Science and Advanced Analytics, DSAA 2023 - Proceedings
BT - 2023 IEEE 10th International Conference on Data Science and Advanced Analytics, DSAA 2023 - Proceedings
A2 - Manolopoulos, Yannis
A2 - Zhou, Zhi-Hua
PB - IEEE
T2 - 10th IEEE International Conference on Data Science and Advanced Analytics, DSAA 2023
Y2 - 9 October 2023 through 12 October 2023
ER -