Learning under Temporal Label Noise

2024-02-06

AI Summary

This paper addresses *temporal label noise*—a previously unformalized challenge in time-series classification where label quality dynamically evolves over time (e.g., gradually improving, deteriorating, or oscillating periodically). We formally define this phenomenon as labels corrupted by a time-dependent noise function. To tackle it, we propose a novel paradigm that abandons the conventional static noise assumption: instead, we directly estimate a time-varying noise transition matrix from data, and integrate it with a robust loss and temporal consistency regularization for noise-aware modeling. Extensive experiments on multiple real-world time-series datasets demonstrate that our method consistently outperforms existing robust learning approaches under diverse temporal noise patterns, achieving state-of-the-art performance. Our core contributions are threefold: (i) the first formal definition and conceptualization of “temporal label noise”; (ii) the first framework for estimating time-varying label noise; and (iii) the first robust classifier explicitly designed for temporal label noise.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Planning, Routing, and Scheduling: Temporal Planning
• Intelligent Robots: State Estimation

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Many time series classification tasks, where labels vary over time, are affected by label noise that also varies over time. Such noise can cause label quality to improve, worsen, or periodically change over time. We first propose and formalize temporal label noise, an unstudied problem for sequential classification of time series. In this setting, multiple labels are recorded over time while being corrupted by a time-dependent noise function. We first demonstrate the importance of modeling the temporal nature of the label noise function and how existing methods will consistently underperform. We then propose methods to train noise-tolerant classifiers by estimating the temporal label noise function directly from data. We show that our methods lead to state-of-the-art performance under diverse types of temporal label noise on real-world datasets

Problem

• Addresses temporal label noise in time series classification.
• Proposes methods to estimate time-dependent label noise functions.
• Improves classifier performance under varying temporal label noise.

Innovation

• Formalizes temporal label noise in time series
• Estimates time-dependent noise function from data
• Develops noise-tolerant classifiers for improved accuracy