Universal Domain Adaptation Benchmark for Time Series Data Representation

2025-05-23

AI Summary

Time-series universal domain adaptation (TS UniDA) lacks a systematic evaluation framework. Method: This paper introduces the UniDA paradigm to time series for the first time and establishes the first standardized benchmark for TS UniDA. We propose a reproducible and extensible evaluation protocol that integrates mainstream backbone architectures—including Informer, TS-TCC, and TS2Vec—and jointly evaluates unsupervised domain adaptation with universal classification loss. Contribution/Results: (1) We reveal the critical influence of backbone selection on generalization and robustness in TS UniDA; (2) Extensive experiments across multiple source datasets demonstrate that our framework significantly improves unknown-class recognition accuracy and cross-domain transfer stability. To foster reproducibility and community advancement, we release the first open-source, standardized evaluation infrastructure for TS UniDA.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Application Domains: Humanities & Computational Social Science
• Multiagent Systems: Adversarial Agents

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Deep learning models have significantly improved the ability to detect novelties in time series (TS) data. This success is attributed to their strong representation capabilities. However, due to the inherent variability in TS data, these models often struggle with generalization and robustness. To address this, a common approach is to perform Unsupervised Domain Adaptation, particularly Universal Domain Adaptation (UniDA), to handle domain shifts and emerging novel classes. While extensively studied in computer vision, UniDA remains underexplored for TS data. This work provides a comprehensive implementation and comparison of state-of-the-art TS backbones in a UniDA framework. We propose a reliable protocol to evaluate their robustness and generalization across different domains. The goal is to provide practitioners with a framework that can be easily extended to incorporate future advancements in UniDA and TS architectures. Our results highlight the critical influence of backbone selection in UniDA performance and enable a robustness analysis across various datasets and architectures.

Problem

• Addressing generalization and robustness in time series data models
• Exploring Universal Domain Adaptation for time series data
• Evaluating backbone robustness in Universal Domain Adaptation frameworks

Innovation

• Unsupervised Domain Adaptation for time series
• Comprehensive comparison of TS backbones
• Robustness protocol across domains