Search papers, labs, and topics across Lattice.
The paper introduces RefQuery, a transfer learning framework for Non-Intrusive Load Monitoring (NILM) that addresses the challenge of cross-domain generalization in appliance energy disaggregation. RefQuery uses compact appliance fingerprints to condition a shared disaggregation model, enabling it to handle multiple appliances without a fixed output set. By freezing a pretrained disaggregation network and learning only per-appliance embeddings during adaptation, RefQuery achieves a strong accuracy-efficiency trade-off, demonstrated on three public datasets.
Forget retraining entire models for each new home – RefQuery's lightweight transfer learning approach lets you disaggregate appliance energy usage by learning only a small embedding per appliance.
Non-Intrusive Load Monitoring (NILM) aims to estimate appliance-level consumption from aggregate electrical signals recorded at a single measurement point. In recent years, the field has increasingly adopted deep learning approaches; however, cross-domain generalization remains a persistent challenge due to variations in appliance characteristics, usage patterns, and background loads across homes. Transfer learning provides a practical paradigm to adapt models with limited target data. However, existing methods often assume a fixed appliance set, lack flexibility for evolving real-world deployments, remain unsuitable for edge devices, or scale poorly for real-time operation. This paper proposes RefQuery, a scalable multi-appliance, multi-task NILM framework that conditions disaggregation on compact appliance fingerprints, allowing one shared model to serve many appliances without a fixed output set. RefQuery keeps a pretrained disaggregation network fully frozen and adapts to a target home by learning only a per-appliance embedding during a lightweight backpropagation stage. Experiments on three public datasets demonstrate that RefQuery delivers a strong accuracy-efficiency trade-off against single-appliance and multi-appliance baselines, including modern Transformer-based methods. These results support RefQuery as a practical path toward scalable, real-time NILM on resource-constrained edge devices.
