COMPGI13 - Advanced Topics in Machine Learning

This database contains the 2016-17 versions of syllabuses. Syllabuses from the 2015-16 session are available here.

Note: Whilst every effort is made to keep the syllabus and assessment records correct, the precise details must be checked with the lecturer(s).

Note: Whilst every effort is made to keep the syllabus and assessment records correct, the precise details must be checked with the lecturer(s).

CodeCOMPGI13 (Also taught as: COMPM050) 
YearMSc
PrerequisitesThe prerequisites are probability, calculus, linear algebra AND COMPGI01 Supervised Learning OR COMPGI08 Graphical Models OR COMPGI18 Probabilistic and Unsupervised Learning
Term2
Taught By

Thore Graepel, Koray Kavukcuoglu, Hado van Hasselt, Joseph Modayil, and additional guest lecturers from Google DeepMind

Aims

To learn the fundamentals of machine-learning based artificial intelligence:

  1. to understand fundamentals and practice of deep learning,
  2. to understand methods for reinforcement learning, planning and control in sequential decision making processes,
  3. to learn how to implement algorithms in TensorFlow.
Learning OutcomesTo gain in depth familiarity with the selected research topics, understand their theory and applications, be able to individually read, understand and discuss research papers in the field.

Content

Introductions to deep learning, reinforcement learning, deep reinforcement learning, and TensorFlow.

  • Review the basics of machine learning, including:

    • linear models
    • least-squares regression
    • maximum likelihood
    • regularization & cross-validation
    • classification
    • logistic regression

  • Deep learning, including:

    • neural networks and convolutional neural networks
    • optimisation
    • end-to-end learning
    • recurrent networks
    • generative models
    • variational inference
    • natural language processing

  • Reinforcement learning, including

    • Markov decision processes
    • dynamic programming
    • model-free prediction and control
    • value function approximation and policy gradients
    • deep reinforcement learning
    • integrating learning and planning
    • exploration and exploitation

Method of Instruction

Lectures, reading, presentation and associated class problems.

Assessment

The course has the following assessment components:

  •     Written Examination (50%)
  •     Coursework Section (50%)

Resources

  • Reinforcement learning:

R. Sutton and A. Barto, An Introduction to Reinforcement Learning

(http://webdocs.cs.ualberta.ca/~sutton/book/ebook/thebook.html)

C. Szepesvari, Algorithms for Reinforcement Learning

(http://www.sztaki.hu/~szcsaba/papers/RLAlgsInMDPslecture.pdf)

  • Deep learning: 

Ian Goodfellow, Yoshua Bengio, and Aaron Courville, Deep Learning

(http://www.deeplearningbook.org/)

  • Tensorflow: 

www.tensorflow.org