New framework SynEnergy generates synthetic energy data preserving anomalies

Researchers have developed SynEnergy, a novel two-stage diffusion-based framework designed to generate synthetic energy consumption data that accurately preserves anomalous events. The framework first uses Heterogeneous Graph-based Anomaly Semantic Learning (HG-ASL) to extract region-specific anomaly semantics by modeling spatial and attribute dependencies. Subsequently, Anomaly Semantic-guided Diffusion (AS-Diff) integrates these learned semantics into the generation process, ensuring that sparse, localized anomalies are maintained alongside realistic consumption patterns. Evaluations on real-world datasets demonstrate SynEnergy's superior fidelity in preserving anomalies and improving downstream task quality compared to existing methods. AI

IMPACT Enables more robust AI applications in energy by providing realistic synthetic data that captures critical anomalous events.

RANK_REASON Academic paper detailing a new method for synthetic data generation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• Anomaly Semantic-guided Diffusion (AS-Diff)
• Heterogeneous Graph-based Anomaly Semantic Learning (HG-ASL)
• SynEnergy

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →