Reinforced Learning - DQN
Published:
This summer I worked as a research intern at Seoul National University Artificial Intelligence Institute’s Biointellignece Lab under the guidance of Professor Byoung-Tak Zhang. During a 3 month research intern period, I primarily conducted research on machine learning, focusing on the areas of reinforced learning (RL) and neural networks. My research topics also included topics such as decision-making under uncertainty, predicting outcomes, and RL in partially observable and noisy environments.
The following presentation is a presentation about Deep Q-Networks (DQN) from a RL seminar hosted by the lab. DQN is a combination of Q-learning, a classic RL algorithm, alongside deep neural networks to handle complex tasks that involve high-dimensional state spaces.
The demo is a Google Colab link to the OpenAI GYM cartpole example included in the presentation.