AI and Kernel Influences: The Impact of Kernel Optimization on Energy Consumption Forecasting

This study addresses the complexities of short-term energy forecasting in the context of fluctuating consumer behaviors and seasonal variations. Our research primarily focused on the development and evaluation of advanced machine learning models and particularly understanding the effects of Kernels on the proposed HyperNetwork-based LSTM model to enhance the accuracy and reliability of energy predictions. We conducted a comprehensive analysis of various state-ofthe-art models, including MLPs, RNNs, and Transformers, rigorously evaluating them across standard metrics for fair comparisons. A significant contribution of our work is the proposed HyperNetwork-based LSTM model which provides consistently superior forecasting accuracy, particularly in certain seasonal contexts. However, upon seasonal evaluation, our proposed solution is not performing well in all seasons. The Python-based implementation of the entire project is available at GitHub Repository — Advanced AI.