"NeuroSwarm: A Hybrid Transformer\u2013Swarm Intelligence Framework for Smart Building Energy Optimization"

"The rising interest in energy-efficient smart buildings necessitates a robust approach to predict and optimize energy consumption. Classical machine learning algorithms and single-layer deep learning techniques fall short when modeling time-series dependencies and optimizing the results. We present a NeuroSwarm (NS-HTS) framework that merges Transformer-based deep learning with swarm intelligence to improve building energy prediction and optimization. The NeuroSwarm framework utilizes Artificial Neural Network (ANN) and CNN-LSTM models to identify non-linear relationships and temporal dependencies, respectively. The Transformer-based deep learning component helps capture long-range dependencies from the ASHRAE Great Energy Predictor III database. On top of that, swarm intelligence plays a key role in optimizing the prediction process by tuning the model hyperparameters. The proposed NeuroSwarm framework provides excellent results in terms of accuracy, root mean square error (RMSE), and mean absolute error (MAE). The framework produces a prediction accuracy of 95.40%, RMSE of 0.162, and MAE of 0.130. In summary, the NeuroSwarm framework outperforms other state-of-the-art methods in predicting building energy consumption and optimizing the output."