Atmospheric Cherenkov telescopes have enabled recent breakthroughs in gamma-ray astronomy, enabling the study of high-energy gamma particles in over 90 galactic and extragalactic regions. The significance of this work arises from the complexity of the data captured by the telescope. Traditional methods may struggle to effectively distinguish between gamma (signal) and hadron (background) events, due to intricate temporal relationships inherent in the data. The dataset used for this research, sourced from the UCI ML repository, simulates the registration of gamma particles. The challenge is to develop a classification model that accurately identifies these gamma events while handling inherent data complexities and normalizing skewed distributions. To address this challenge, a classification model is developed using ten features from the MAGIC gamma telescope dataset. This research introduces the innovative application of deep learning techniques, specifically Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Bidirectional LSTM (Bi-LSTM), to the field of gamma-ray astronomy to classify high-energy gamma particles detected by the Atmospheric Cherenkov telescopes. Furthermore, the research introduces the application of square root transformation as a method to address skewness and kurtosis in the dataset. This preprocessing technique aids in normalizing data distributions, which is crucial for accurate model training and classification. By leveraging the power of deep learning and innovative data transformations, the best accuracy of 88.71% is achieved by the LSTM+ReLU model with three layers for gamma and hadron particle classification. These findings offer insights into fundamental astrophysical processes and contribute to the advancement of gamma-ray astronomy.