 Deep learning techniques have shown great promise in recent years, however their energy consumption and carbon footprint are often overlooked. This paper proposes a novel approach to reduce the environmental costs associated with deep learning models by employing a stochastic hyperparameter selection technique. Experiments on various datasets and models demonstrate that the proposed approach can achieve similar accuracy levels to existing methods while significantly reducing energy consumption and CO2 emissions. This article was authored by Kazir Afat, Saadia Islam, Abdullah al-Mafiq, and others.