Imbalance-aware method improves forecasting of rare time patterns
IMRouter: Imbalance-aware Temporal Pattern Routing for LLM-based Time Series Forecasting
Machine Learning
Summary
Sometimes, computers that predict future events like traffic or online resource needs make more mistakes when dealing with rare patterns because they focus too much on common situations. The paper’s authors identified this problem and created a new method called IMRouter, which helps the computer pay more attention to those rare patterns. Their method uses clever ways to teach different model parts to specialize and share knowledge about unusual patterns. Tests showed that IMRouter improves accuracy, especially for those less frequent but important patterns.
What this means in practice
- •For traffic management teams: Enhance traffic flow predictions by improving model accuracy on rare traffic scenarios using IMRouter’s imbalance-aware techniques.
- •For cloud resource managers: Boost online resource allocation forecasts by better handling infrequent usage patterns through imbalance-aware time series models.
Authors
Xin Yang, Cai Xu, Haishun Chen, Ying Liu, Jie Zhao, Wei Zhao, Ziyu Guan
Abstract
Accurate time series forecasting (TSF) is essential for Web applications, such as traffic monitoring and online resource management. Recent advances in leveraging pre-trained large language models (LLMs) have shown promising performance in TSF. However, existing LLM-based TSF methods largely optimize the overall forecasting objective while overlooking temporal pattern imbalance (TPI), where different temporal patterns occur at highly unequal frequencies. Our empirical analysis reveals that TPI is prevalent across widely used TSF benchmarks, with substantially higher forecasting errors on less frequent patterns. To address this issue, we propose IMRouter (Imbalance-aware Temporal Pattern Routing), a novel framework that mitigates TPI-induced biases in both temporal representation learning and cross-modal alignment. Specifically, IMRouter introduces Pattern-Adaptive Temporal Representation Learning (PATR), which leverages gradient conflict and local gradient density to guide expert specialization. By penalizing incompatible expert assignments with density-aware weighting, PATR reduces the dominance of frequent optimization signals in router training. Furthermore, Routing-Guided Semantic Alignment (RGSA) transfers the pattern-specific knowledge learned by the router to cross-modal alignment through mutual information maximization, preserving distinctive temporal information across different patterns. Extensive experiments on benchmark datasets demonstrate the effectiveness of IMRouter, particularly in improving forecasting performance on less frequent temporal patterns.