Family Matters: A Systematic Study of Spatial vs. Frequency Masking for Continual Test-Time Adaptation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Doloriel, Chandler Timm C., Zhang, Yunbei, Yu, Yeonguk, Rafi, Taki Hasan, siddiqui, Muhammad salman, Stevik, Tor Kristian, Machot, Fadi Al, Liland, Kristian Hovde, Ullah, Habib
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914619982872576
author Doloriel, Chandler Timm C.
Zhang, Yunbei
Yu, Yeonguk
Rafi, Taki Hasan
siddiqui, Muhammad salman
Stevik, Tor Kristian
Machot, Fadi Al
Liland, Kristian Hovde
Ullah, Habib
author_facet Doloriel, Chandler Timm C.
Zhang, Yunbei
Yu, Yeonguk
Rafi, Taki Hasan
siddiqui, Muhammad salman
Stevik, Tor Kristian
Machot, Fadi Al
Liland, Kristian Hovde
Ullah, Habib
contents Recent continual test-time adaptation (CTTA) methods adopt masked image modeling to stabilize learning under distribution shift, yet each treats its masking family F as a fixed design choice and innovates exclusively along the selection strategy S, leaving the family axis underexplored. We present a systematic empirical study that isolates this axis. Using a controlled CTTA instantiation -- Mask to Adapt (M2A) -- that fixes S = random and standard losses, we vary only F across spatial (patch, pixel) and frequency (all-band, low-band, high-band) families while keeping every other component identical. The study's contributions are the design guidance it extracts for the CTTA settings we evaluated: (1) the masking family determines whether adaptation compounds useful structure or compounds errors -- on patch-tokenized architectures, spatial masking accumulates stable representations over long streams while frequency masking collapses catastrophically. We characterize this instability through a structural-preservation account, where spatial coherence maintains the broad-spectrum redundancy needed to avoid terminally overlapping with a corruption's spectral signature; (2) the optimal family depends on architecture-task alignment -- on CNNs, whose overlapping receptive fields dilute patch occlusion, the family gap vanishes, whereas on fine-grained tasks with global cues and large-capacity ViTs, frequency masking becomes competitive. In confounded system-level comparisons -- where baselines also differ in losses and auxiliary components -- M2A's random selection performs comparably to heuristic strategies, though we treat this observation as suggestive context rather than a controlled quantification of S's relative importance.
format Preprint
id arxiv_https___arxiv_org_abs_2512_08048
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Family Matters: A Systematic Study of Spatial vs. Frequency Masking for Continual Test-Time Adaptation
Doloriel, Chandler Timm C.
Zhang, Yunbei
Yu, Yeonguk
Rafi, Taki Hasan
siddiqui, Muhammad salman
Stevik, Tor Kristian
Machot, Fadi Al
Liland, Kristian Hovde
Ullah, Habib
Computer Vision and Pattern Recognition
Recent continual test-time adaptation (CTTA) methods adopt masked image modeling to stabilize learning under distribution shift, yet each treats its masking family F as a fixed design choice and innovates exclusively along the selection strategy S, leaving the family axis underexplored. We present a systematic empirical study that isolates this axis. Using a controlled CTTA instantiation -- Mask to Adapt (M2A) -- that fixes S = random and standard losses, we vary only F across spatial (patch, pixel) and frequency (all-band, low-band, high-band) families while keeping every other component identical. The study's contributions are the design guidance it extracts for the CTTA settings we evaluated: (1) the masking family determines whether adaptation compounds useful structure or compounds errors -- on patch-tokenized architectures, spatial masking accumulates stable representations over long streams while frequency masking collapses catastrophically. We characterize this instability through a structural-preservation account, where spatial coherence maintains the broad-spectrum redundancy needed to avoid terminally overlapping with a corruption's spectral signature; (2) the optimal family depends on architecture-task alignment -- on CNNs, whose overlapping receptive fields dilute patch occlusion, the family gap vanishes, whereas on fine-grained tasks with global cues and large-capacity ViTs, frequency masking becomes competitive. In confounded system-level comparisons -- where baselines also differ in losses and auxiliary components -- M2A's random selection performs comparably to heuristic strategies, though we treat this observation as suggestive context rather than a controlled quantification of S's relative importance.
title Family Matters: A Systematic Study of Spatial vs. Frequency Masking for Continual Test-Time Adaptation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2512.08048