Adaptive Multi-task Learning for Probabilistic Load Forecasting

2025-12-23

AI Summary

Addressing the challenges of modeling dynamically evolving consumption patterns, time-varying inter-entity correlations, and uncertainty quantification in multi-region/multi-building probabilistic load forecasting, this paper proposes the first online adaptive multitask probabilistic forecasting framework. Innovatively integrating an adaptive mechanism into multitask load prediction, the method jointly leverages vector-valued hidden Markov models (VV-HMMs) and recursive Bayesian parameter updating to simultaneously learn shared representations across entities and model dynamic correlations. Departing from conventional offline paradigms, the framework enables real-time parameter adaptation and uncertainty calibration. Evaluated on a multi-source dynamic load dataset, it achieves a 12.7% improvement in predictive accuracy, a 19.3% reduction in quantile loss, and an uncertainty calibration score of 92.1%.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Relational Probabilistic Models
• Intelligent Robots: State Estimation

Application Category

• User Modeling, Personalization and Recommendation: User modeling for targeted and personalized online advertising
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Simultaneous load forecasting across multiple entities (e.g., regions, buildings) is crucial for the efficient, reliable, and cost-effective operation of power systems. Accurate load forecasting is a challenging problem due to the inherent uncertainties in load demand, dynamic changes in consumption patterns, and correlations among entities. Multi-task learning has emerged as a powerful machine learning approach that enables the simultaneous learning across multiple related problems. However, its application to load forecasting remains underexplored and is limited to offline learning-based methods, which cannot capture changes in consumption patterns. This paper presents an adaptive multi-task learning method for probabilistic load forecasting. The proposed method can dynamically adapt to changes in consumption patterns and correlations among entities. In addition, the techniques presented provide reliable probabilistic predictions for loads of multiples entities and assess load uncertainties. Specifically, the method is based on vectorvalued hidden Markov models and uses a recursive process to update the model parameters and provide predictions with the most recent parameters. The performance of the proposed method is evaluated using datasets that contain the load demand of multiple entities and exhibit diverse and dynamic consumption patterns. The experimental results show that the presented techniques outperform existing methods both in terms of forecasting performance and uncertainty assessment.

Problem

• Adaptive multi-task learning for probabilistic load forecasting
• Dynamic adaptation to changing consumption patterns and correlations
• Reliable probabilistic predictions and uncertainty assessment for multiple entities

Innovation

• Adaptive multi-task learning for dynamic pattern adaptation
• Vector-valued hidden Markov models for probabilistic forecasting
• Recursive parameter updates for real-time prediction accuracy