{"ID":2865080,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2509.21934","arxiv_id":"2509.21934","title":"Extracting Actionable Insights from Building Energy Data using Vision LLMs on Wavelet and 3D Recurrence Representations","abstract":"The analysis of complex building time-series for actionable insights and recommendations remains challenging due to the nonlinear and multi-scale characteristics of energy data. To address this, we propose a framework that fine-tunes visual language large models (VLLMs) on 3D graphical representations of the data. The approach converts 1D time-series into 3D representations using continuous wavelet transforms (CWTs) and recurrence plots (RPs), which capture temporal dynamics and localize frequency anomalies. These 3D encodings enable VLLMs to visually interpret energy-consumption patterns, detect anomalies, and provide recommendations for energy efficiency. We demonstrate the framework on real-world building-energy datasets, where fine-tuned VLLMs successfully monitor building states, identify recurring anomalies, and generate optimization recommendations. Quantitatively, the Idefics-7B VLLM achieves validation losses of 0.0952 with CWTs and 0.1064 with RPs on the University of Sharjah energy dataset, outperforming direct fine-tuning on raw time-series data (0.1176) for anomaly detection. This work bridges time-series analysis and visualization, providing a scalable and interpretable framework for energy analytics.","short_abstract":"The analysis of complex building time-series for actionable insights and recommendations remains challenging due to the nonlinear and multi-scale characteristics of energy data. To address this, we propose a framework that fine-tunes visual language large models (VLLMs) on 3D graphical representations of the data. The...","url_abs":"https://arxiv.org/abs/2509.21934","url_pdf":"https://arxiv.org/pdf/2509.21934v1","authors":"[\"Amine Bechar\",\"Adel Oulefki\",\"Abbes Amira\",\"Fatih Kurogollu\",\"Yassine Himeur\"]","published":"2025-09-26T06:18:51Z","proceeding":"cs.LG","tasks":"[\"cs.LG\",\"cs.CY\"]","methods":"[\"Large Language Model\"]","has_code":false}