An Enhanced Optimization Technique to Implement Dino Game using Machine Learning Techniques

Abstract

In this task, we put in force both characteristic-extraction based algorithms and a give up-to-cease deep reinforcement gaining knowledge of approach to learning to manipulate Chrome offline dinosaur recreation at once from excessive-dimensional recreation display screen input. Results show that as compared with the pixel feature primarily based algorithms, deep reinforcement learning is extra powerful and powerful. It leverages the high-dimensional sensory enter at once and avoids capacity errors in feature extraction. Finally, we advise unique schooling methods to address magnificence imbalance issues caused by the boom in-game speed. After education, our Deep-Q AI can outperform human specialists.