Predictive Spectral Calibration for Source-Free Test-Time Regression

TL;DR

This paper introduces Predictive Spectral Calibration (PSC), a source-free method for test-time regression that improves adaptation to distribution shifts by spectral feature alignment, demonstrating consistent gains across benchmarks.

Contribution

PSC extends subspace alignment to spectral matching, enabling source-free, model-agnostic test-time adaptation for image regression tasks.

Findings

PSC outperforms strong baselines under severe distribution shifts.

The method is simple, model-agnostic, and compatible with pretrained regressors.

Experiments show consistent improvements across multiple benchmarks.

Abstract

Test-time adaptation (TTA) for image regression has received far less attention than its classification counterpart. Methods designed for classification often depend on classification-specific objectives and decision boundaries, making them difficult to transfer directly to continuous regression targets. Recent progress revisits regression TTA through subspace alignment, showing that simple source-guided alignment can be both practical and effective. Building on this line of work, we propose Predictive Spectral Calibration (PSC), a source-free framework that extends subspace alignment to block spectral matching. Instead of relying on a fixed support subspace alone, PSC jointly aligns target features within the source predictive support and calibrates residual spectral slack in the orthogonal complement. PSC remains simple to implement, model-agnostic, and compatible with off-the-shelf pretrained regressors. Experiments on multiple image regression benchmarks show consistent improvements over strong baselines, with particularly clear gains under severe distribution shifts.