Amortized Predictability-aware Training Framework for Time Series Forecasting and Classification

2026-02-18

AI Summary

This work addresses the challenge posed by low-predictability samples in time series data, which often lead to unstable training dynamics or convergence to suboptimal solutions. To mitigate this issue, the authors propose a general Amortized Predictability-aware Training Framework (APTF) that dynamically identifies low-predictability instances and adaptively reweights their contribution to the loss. The framework incorporates a Hierarchical Predictability-aware Loss (HPL), which progressively intensifies penalties on hard-to-predict samples during training, and an amortized predictability estimation module that effectively reduces estimation errors caused by model bias. Empirical results demonstrate that the proposed approach significantly enhances both performance and training stability across time series forecasting and classification tasks.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Reasoning under Uncertainty: Stochastic Optimization

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Time series data are prone to noise in various domains, and training samples may contain low-predictability patterns that deviate from the normal data distribution, leading to training instability or convergence to poor local minima. Therefore, mitigating the adverse effects of low-predictability samples is crucial for time series analysis tasks such as time series forecasting (TSF) and time series classification (TSC). While many deep learning models have achieved promising performance, few consider how to identify and penalize low-predictability samples to improve model performance from the training perspective. To fill this gap, we propose a general Amortized Predictability-aware Training Framework (APTF) for both TSF and TSC. APTF introduces two key designs that enable the model to focus on high-predictability samples while still learning appropriately from low-predictability ones: (i) a Hierarchical Predictability-aware Loss (HPL) that dynamically identifies low-predictability samples and progressively expands their loss penalty as training evolves, and (ii) an amortization model that mitigates predictability estimation errors caused by model bias, further enhancing HPL's effectiveness. The code is available at https://github.com/Meteor-Stars/APTF.

Problem

• time series forecasting
• time series classification
• low-predictability samples
• training instability
• noise

Innovation

• Amortized Predictability-aware Training
• Hierarchical Predictability-aware Loss
• Time Series Forecasting
• Time Series Classification
• Low-predictability Sample Mitigation