COM3240 Adaptive Intelligence
||This course will examine the theme of bio-inspired Machine Learning. The students will be taught supervised, unsupervised and reinforcement learning methods in Neural Networks, including state-of-the-art algorithms (Deep Reinforcement Learning), as well as their underpinning mathematical principles and biological inspiration.
||Dr Matthew Ellis
||To teach students the theory and
implementation of bio-inspired machine learning algorithms.
||By the end of the course, students should:
- understand the theory, implementation and application
of supervised, unsupervised and reinforcement learning algorithms
- be able to read relevant textbooks and research
- be able to write research reports
||Topics to include:
- Supervised learning: the backpropagation algorithm.
- Learning and Memory in Brain Circuits and Artificial Neural Networks
- Unsupervised Learning
(e.g. Oja's rule/Principal Component Analysis, Clustering/Competitive Learning)
- Reinforcement Learning
(e.g. Temporal Difference Learning: Q learning, SARSA, and Deep Reinforcement Learning).
|Students from departments other than Computer Science will
need to demonstrate a good understanding of programming and
||The course will be delivered as a series of pre-recorded video lectures that students are required to watch prior to a
weekly 1 hour interactive session and a 1 hour practical lab session. These will be backed up by exercises and
homework. Group discussion of issues relevant to the course
will be encouraged. Self-study of the course is also required, supported by a variety of resources (e.g. journal articles, books). Knowledge of programming (Matlab or Python) and A level maths is compulsory.
||Students will receive feedback on assignments in writing via Blackboard and via an in-class feedback session. Feedback will also be delivered during labs and office hours (upon request).
A selection from: