Mitigating Missing Feature Channels at Inference Stage: Test-Time Adaptation Through Self-Training With Data Imputation

Page view(s)
0
Checked on
Mitigating Missing Feature Channels at Inference Stage: Test-Time Adaptation Through Self-Training With Data Imputation
Title:
Mitigating Missing Feature Channels at Inference Stage: Test-Time Adaptation Through Self-Training With Data Imputation
Journal Title:
IEEE Signal Processing Letters
Keywords:
Publication Date:
14 February 2025
Citation:
Huang, Y., Su, Y., Yang, X., Lin, X., & Xu, X. (2025). Mitigating Missing Feature Channels at Inference Stage: Test-Time Adaptation Through Self-Training With Data Imputation. IEEE Signal Processing Letters, 32, 2414–2418. https://doi.org/10.1109/lsp.2025.3542696
Abstract:
The robustness of deep learning model can be compromised by out-of-distribution (OOD) testing data. Test-time adaptation (TTA) emerges as an efficient method to mitigate the distribution gap by tuning model weights at inference stage. TTA are mainly demonstrated on robustifying model on additive visual corruptions or adversarial attacks. In this work, we specify an overlooked type of OOD where feature channels could be missing in testing data, potentially due to sensor fault. We reveal that self-training and data imputation can improve the model’s generalization to data with missing feature channel. To address the uncertainty associated with imputed samples, we fuse predictions from imputed and weakly-augmented samples for more reliable pseudo labels. We evaluate the effectiveness on multiple image classification benchmarks with synthesized and realistic missing feature channels, and our proposed method outperforms state-of-the-art TTA methods on all benchmarks.
License type:
Publisher Copyright
Funding Info:
This research / project is supported by the A*STAR - Manufacturing, Trade, and Connectivity Programmatic Fund
Grant Reference no. : M23L7b0021
Description:
© 2025 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
ISSN:
1070-9908
1558-2361
Files uploaded:

File Size Format Action
astar-oar.pdf 685.48 KB PDF Open