COMP3667: REINFORCEMENT LEARNING

• (COMP2261 Artificial Intelligence AND COMP2271 Data Science)

• COMP3547 Deep Learning

• None

• To understand computational models of learning in dynamic environments.
• Learning how to plan and design agents with intelligent behaviour, taking actions to control the environment and maximise cumulative future rewards.

• Introduction to reinforcement learning.
• Markov decision processes and planning by dynamic programming.
• Model free prediction and control.
• Value-based and policy-based reinforcement learning.
• Scaling up reinforcement learning approaches with deep learning.
• Integrating learning and planning, and balancing exploration/exploitation.

Subject-specific Knowledge:

• On completion of the module, students will be able to demonstrate:
• an understanding of the key features of reinforcement learning and differences with non-interactive learning.
• an understanding of state-of-the-art reinforcement learning algorithms.
• an understanding of the issues faced in scaling reinforcement learning approaches using deep learning.

Subject-specific Skills:

• On completion of the module, students will be able to demonstrate:
• an ability to use modern libraries to design, train, validate and test deep reinforcement learning models.
• an ability to find RL based solutions with respect to the task or environment.
• an ability to design bespoke RL algorithms based on the problem and the environment, such as whether in continuous or discrete action spaces.
• an ability to solve complex learning and planning problems in dynamic environments.

Key Skills:

• On completion of the module, students will be able to demonstrate:
• the scientific approach to the design, training, validation, and testing of reinforcement techniques in a broad range of applications.
• an ability to design new environments with OpenAI gym, and design tailored agents that learn to control the environments.
• an ability to identify the problem area and subsequently design and implement state-of-the-art reinforcement learning approaches.

• Lectures enable the students to learn new material relevant to reinforcement learning, as well as its applications.
• Practicals enable students to acquire the necessary coding skills, learn about the relevant libraries and packages and receive feedback on their work.
• Summative assessments assess the knowledge of relevant libraries and application of methods and techniques.

Example formative exercises are given during the course. The first few lab practicals are dedicated to formative assignments aimed at familiarising students with state-of-the-art packages and libraries used in reinforcement learning. Feedback will be provided to the students on the summative assignments and lecture materials during the practicals. Additional revision lectures may be arranged in the module's lecture slots in the 3rd term.

If you have a question about Durham's modular degree programmes, please visit our FAQ webpages, Help page or our glossary of terms. If you have a question about modular programmes that is not covered by the FAQ, or a query about the on-line Undergraduate Module Handbook, please contact us.

Prospective Students: If you have a query about a specific module or degree programme, please Ask Us.

Current Students: Please contact your department.