Efficient Test-Time Adaptation through Latent Subspace Coefficients Search
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866914535970963456 |
|---|---|
| author | Luo, Xinyu Liu, Jie Chen, Kecheng Yang, Junyi Ding, Bo Basu, Arindam Li, Haoliang |
| author_facet | Luo, Xinyu Liu, Jie Chen, Kecheng Yang, Junyi Ding, Bo Basu, Arindam Li, Haoliang |
| contents | Real-world deployment often exposes models to distribution shifts, making test-time adaptation (TTA) critical for robustness. Yet most TTA methods are unfriendly to edge deployment, as they rely on backpropagation, activation buffering, or test-time mini-batches, leading to high latency and memory overhead. We propose \textbf{ELaTTA} (\textit{Efficient Latent Test-Time Adaptation}), a gradient-free framework for single-instance TTA under strict on-device constraints. ELaTTA freezes model weights and adapts each test sample by optimizing a low-dimensional coefficient vector in a source-induced principal latent subspace, pre-computed offline via truncated SVD and stored with negligible overhead. At inference, ELaTTA encourages prediction confidence by optimizing the $k$-D coefficients with CMA-ES, effectively optimizing a Gaussian-smoothed objective and improving stability near decision boundaries. Across six benchmarks and multiple architectures, ELaTTA achieves state-of-the-art accuracy under both strict and continual single-instance protocols, while reducing compute by up to \emph{63$\times$} and peak memory by up to \emph{11$\times$}. We further demonstrate on-device deployment on a ZYNQ-7020 platform. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_11068 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Efficient Test-Time Adaptation through Latent Subspace Coefficients Search Luo, Xinyu Liu, Jie Chen, Kecheng Yang, Junyi Ding, Bo Basu, Arindam Li, Haoliang Machine Learning Audio and Speech Processing Image and Video Processing Real-world deployment often exposes models to distribution shifts, making test-time adaptation (TTA) critical for robustness. Yet most TTA methods are unfriendly to edge deployment, as they rely on backpropagation, activation buffering, or test-time mini-batches, leading to high latency and memory overhead. We propose \textbf{ELaTTA} (\textit{Efficient Latent Test-Time Adaptation}), a gradient-free framework for single-instance TTA under strict on-device constraints. ELaTTA freezes model weights and adapts each test sample by optimizing a low-dimensional coefficient vector in a source-induced principal latent subspace, pre-computed offline via truncated SVD and stored with negligible overhead. At inference, ELaTTA encourages prediction confidence by optimizing the $k$-D coefficients with CMA-ES, effectively optimizing a Gaussian-smoothed objective and improving stability near decision boundaries. Across six benchmarks and multiple architectures, ELaTTA achieves state-of-the-art accuracy under both strict and continual single-instance protocols, while reducing compute by up to \emph{63$\times$} and peak memory by up to \emph{11$\times$}. We further demonstrate on-device deployment on a ZYNQ-7020 platform. |
| title | Efficient Test-Time Adaptation through Latent Subspace Coefficients Search |
| topic | Machine Learning Audio and Speech Processing Image and Video Processing |
| url | https://arxiv.org/abs/2510.11068 |