Energy Disaggregation Using Radial Basis Function Neural Networks based on Deep Co Training Architecture
Bibliographic record
Abstract
Non-Intrusive Load Monitoring (NILM) has emerged as a critical tool in energy management, offering a non- invasive means to disaggregate a building’s total electricity usage into appliance-level consumption. However, existing NILM approaches often require large labeled datasets, raising significant challenges in terms of dataset availability, privacy exposure, and computational burdens. This paper proposes a semi-supervised energy disaggregation approach using a Radial Basis Function Neural Network (RBFNN) within a Deep Co-Training framework. RBFNN’s fast convergence and robustness to noise address key practical issues in NILM research, such as limited labeling and the complexity of real-world power signals. Meanwhile, the co-training paradigm allows two neural networks to iteratively exchange pseudo-labels on unlabeled data, thus significantly reducing the need for extensive ground-truth annotations. We validate our approach on two public datasets, REDD and UK-DALE, disaggregating six household appliances. Experimental results demonstrate superior performance with improved Accuracy, Precision, Recall, and F1-score when compared with three state-of-the- art semi-supervised NILM methods. Our work underscores the utility of combining RBFNNs with deep co-training to address data scarcity, computational efficiency, and privacy concerns, ultimately facilitating broader deployment of NILM solutions for energy conservation and sustainability.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".