A Comprehensive Benchmark of Source-Free Universal Domain Adaptation on Time Series Representations

2026-09-30

AI Summary

This study addresses the absence of source-free universal domain adaptation (SF-UniDA) benchmarks for time series data and the high sensitivity of existing methods to rejection thresholds for unknown samples. To this end, we construct the first time series SF-UniDA benchmark and systematically evaluate the feature extraction capabilities of pretrained foundation models. Furthermore, we propose a plug-and-play automatic thresholding module to optimize the handling of unknown samples. Experimental results validate the effectiveness of the proposed module. Notably, our findings reveal that foundation models do not systematically outperform classical backbone networks in this context, thereby underscoring the necessity of developing time-series-specific SF-UniDA methodologies.

Abstract

Source-Free Universal Domain Adaptation (SF-UniDA) extends Universal Domain Adaptation by removing access to source data at adaptation time while still handling label-set mismatches between domains. Despite growing interest in this setting for image data, no benchmark exists for time series, which are more challenging. We present the first SF-UniDA benchmark on time series. In addition, we provide the first study of pretrained foundation models as feature extractors for time series domain adaptation. In this context, we identify a critical and previously underexplored limitation of all existing SF-UniDA methods: the inference threshold for unknown-sample rejection is highly sensitive. We address this by proposing a plug-in auto-thresholding module that can be integrated into any SF-UniDA method. Experiments on three well-known time series datasets confirm the suitability of this module. They also highlight that foundation models do not systematically outperform classical backbones and that SF-UniDA tailored for time series is yet to be developed.

Problem

• Source-Free Universal Domain Adaptation
• Time Series
• Benchmark
• Unknown-sample rejection
• Threshold sensitivity

Innovation

• Source-Free Universal Domain Adaptation
• Time Series
• Foundation Models
• Auto-thresholding Module
• Benchmark