CaberNet: Causal Representation Learning for Cross-Domain HVAC Energy Prediction

2025-11-10

AI Summary

Cross-domain HVAC energy consumption forecasting suffers from data scarcity, strong heterogeneity across building environments, and spurious correlations that induce overfitting. To address these challenges, we propose CaberNet—the first causal representation learning framework specifically designed for HVAC energy prediction. CaberNet identifies causal features via global feature gating and self-supervised Bernoulli regularization, while enforcing domain-invariant Markov blanket representations through domain-balanced training and latent factor independence constraints—without requiring expert priors or strong structural assumptions. Integrating deep sequential modeling with self-supervised learning, CaberNet is evaluated on three real-world building datasets spanning markedly distinct climate zones. It achieves a 22.9% reduction in normalized mean squared error over the best baseline, demonstrating substantial improvements in cross-domain generalization and model interpretability.

Technology Category

• Machine Learning: Causal Learning
• Computer Vision: Representation Learning for Vision
• Knowledge Representation and Reasoning: Action, Change, and Causality

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Cross-domain HVAC energy prediction is essential for scalable building energy management, particularly because collecting extensive labeled data for every new building is both costly and impractical. Yet, this task remains highly challenging due to the scarcity and heterogeneity of data across different buildings, climate zones, and seasonal patterns. In particular, buildings situated in distinct climatic regions introduce variability that often leads existing methods to overfit to spurious correlations, rely heavily on expert intervention, or compromise on data diversity. To address these limitations, we propose CaberNet, a causal and interpretable deep sequence model that learns invariant (Markov blanket) representations for robust cross-domain prediction. In a purely data-driven fashion and without requiring any prior knowledge, CaberNet integrates i) a global feature gate trained with a self-supervised Bernoulli regularization to distinguish superior causal features from inferior ones, and ii) a domain-wise training scheme that balances domain contributions, minimizes cross-domain loss variance, and promotes latent factor independence. We evaluate CaberNet on real-world datasets collected from three buildings located in three climatically diverse cities, and it consistently outperforms all baselines, achieving a 22.9% reduction in normalized mean squared error (NMSE) compared to the best benchmark. Our code is available at https://github.com/rickzky1001/CaberNet-CRL.

Problem

• Addresses cross-domain HVAC energy prediction challenges due to data scarcity and heterogeneity
• Mitigates overfitting to spurious correlations across different buildings and climates
• Reduces reliance on expert intervention while maintaining data diversity

Innovation

• Learns invariant causal representations for cross-domain prediction
• Uses global feature gate with self-supervised Bernoulli regularization
• Implements domain-wise training scheme balancing domain contributions