Augmented Contrastive Clustering with Uncertainty-Aware Prototyping for Time Series Test Time Adaptation

2025-01-01

AI Summary

Existing test-time adaptation (TTA) methods exhibit weak dynamic modeling and insufficient uncertainty handling on complex real-world time-series data (e.g., video, audio). To address this, we propose the first TTA framework specifically designed for sequential data. Our method operates label-free and online, requiring no architectural modifications to arbitrary pre-trained models. Key contributions include: (1) uncertainty-aware prototype distillation, which explicitly models temporal prediction confidence via calibrated uncertainty estimation; and (2) an entropy-driven pseudo-label selection mechanism jointly optimized with contrastive clustering, effectively suppressing noise accumulation while promoting intra-class compactness and inter-class separability in feature space. Extensive experiments on three real-world time-series benchmarks demonstrate significant improvements over state-of-the-art TTA approaches. Cross-domain evaluation further confirms strong generalization capability and architectural agnosticism, validating broad applicability across vision tasks.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Sequential Decision Making
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling

Abstract

Test-time adaptation aims to adapt pre-trained deep neural networks using solely online unlabelled test data during inference. Although TTA has shown promise in visual applications, its potential in time series contexts remains largely unexplored. Existing TTA methods, originally designed for visual tasks, may not effectively handle the complex temporal dynamics of real-world time series data, resulting in suboptimal adaptation performance. To address this gap, we propose Augmented Contrastive Clustering with Uncertainty-aware Prototyping (ACCUP), a straightforward yet effective TTA method for time series data. Initially, our approach employs augmentation ensemble on the time series data to capture diverse temporal information and variations, incorporating uncertainty-aware prototypes to distill essential characteristics. Additionally, we introduce an entropy comparison scheme to selectively acquire more confident predictions, enhancing the reliability of pseudo labels. Furthermore, we utilize augmented contrastive clustering to enhance feature discriminability and mitigate error accumulation from noisy pseudo labels, promoting cohesive clustering within the same class while facilitating clear separation between different classes. Extensive experiments conducted on three real-world time series datasets and an additional visual dataset demonstrate the effectiveness and generalization potential of the proposed method, advancing the underexplored realm of TTA for time series data.

Problem

• Time Series Analysis
• Temporal Information Capture
• Uncertainty Handling

Innovation

• Enhanced Contrastive Clustering
• Uncertainty-aware Prototypes
• Time Series Test-Time Adaptation