Oxford, United Kingdom

Advanced Artificial Intelligence and Machine Learning: Reinforcement Learning

Getting things wrong is part of what makes us human, and our natural intelligence helps us learn from our mistakes. Reinforcement learning is an area of machine learning which enables artificial intelligence to learn from its mistakes as well, for example allowing a robot to use trial-and-error to interact with a new environment and achieve an objective. This advanced course examines the fundamentals of reinforcement learning and explores the varied applications of dynamic programming methods.

The course will begin with a thorough grounding in the key theoretical concepts of reinforcement learning, familiarising you with agents, environments, and rewards, before introducing Markov decision processes, dynamic programming, and Monte Carlo methods. As the course progresses you will explore a wide range of reinforcement learning methods and techniques, including policy gradient methods and how they optimise policies, policy search methods such as evolutionary strategies and hill-climbing, and the cross-entropy method for policy optimisation. The final part of the course will introduce even more advanced topics, including multi-agent reinforcement learning.

This intensive course offers students theoretical understanding and practical experience in a range of reinforcement learning concepts and techniques, offering career skills as well as excellent foundations for future research.

Target group

This course would suit STEM students with intermediate level experience in artificial intelligence and machine learning concepts and techniques, including those undertaking, or looking ahead to, graduate level study or research.

Specifically, students on this course must have experience of the following topics:
- Knowledge of the deep learning libraries.
- Understanding of deep learning, neural networks and basic dynamic programming.
- Strong background in optimization and probability.
- Familiarity with the Python programming language.

Course aim

By the end of this course, you will:
- Understand the fundamentals of reinforcement learning, including agents, environments, and rewards.
- Be able to assess and utilise a range of reinforcement learning approaches.
- Be able to evaluate the efficacy of a range of reinforcement learning methods.
- Understand different strategies for training multiple agents, both decentralised and centralised.
- Demonstrate familiarity with current research.

Credits info

7.5 EC
LMH Summer Programmes are designed to be eligible for credit, and we recommend the award of 7.5 ECTS / 4 US / 15 CATS for this course.

Fee info

GBP 1360: This includes:
- All tuition, including lectures, seminars, and tutorials.
- Assessment, transcript of academic performance, and certificate.
- Access to the LMH Summer Programmes remote learning platform.
- Support of the dedicated Remote Learning Coordinator.