Description:
This project addresses the challenge of optimizing resource allocation on live-streaming video platforms.
It proposes a sophisticated approach that combines Machine Learning-Deep Q-Learning (ML-DQL), Double Deep Q-Network (Double DQN),
and Count-based Exploration DQN. The focus is on solving the problem of efficiently distributing server resources amidst the dynamically
changing demand of users, ensuring both high-quality streaming experience and cost-effectiveness. The paper’s findings demonstrate
significant improvements in resource utilization, offering valuable insights for managing complex, user-driven live-streaming environments.
Report: Please mail to me (hw2894@columbia.edu) or through the mail symbol at the right bottom.