Branch prediction is a crucial component of modern micro-processors, playing a pivotal role in enhancing instruction-level parallelism and overall processor performance. Accurate branch prediction reduces pipeline stalls, ensuring efficient execution of instructions. Traditional branch prediction methods, such as static and dynamic predictors, have made substantial progress but still exhibit limitations in handling complex and irregular branch patterns. This research explores the application of deep learning and machine learning models, coupled with architectural features, to improve branch prediction accuracy by leveraging the power of neural networks and advanced machine learning techniques. In this paper, our primary emphasis is on traditional branch prediction algorithms. We critically examine their limitations and present a comprehensive literature survey on the integration of deep learning techniques to develop dynamic branch predictors capable of accurately forecasting conditional branch instructions. While previous surveys in this domain have concentrated on dynamic branch prediction techniques relying on neural network perceptions, we intend to enhance and expand upon these surveys by incorporating the most recent advancements in deep learning (DL) and advanced machine learning (ML)-based branch prediction methods.