CSC-M37 Data Visualisation at Swansea University
Data Visualization is concerned with the automatic or semi-automatic generation of digital images that depict data in meaningful ways. It is a relatively new field of computer science that is rapidly evolving and expanding. It is also very application-oriented, i.e., real tools are built in order to help scientists from other disciplines. Students will be encouraged incorporate data from their own research domains in the exercises.
The course covers three strands:
Information visualisation: including abstract vs. hierarchical data, tree maps, cone trees, focus and context techniques, multi-dimensional data, scatter plots, icons, parallel coordinates, interaction techniques, linking and brushing.
Volume visualisation: covering slicing, surface vs. volume rendering, transfer functions, interpolation schemes, direct volume visualisation, ray casting, isosurfacing.
Flow visualisation: simulation, steady and time-dependent flow, direct and indirect flow visualisation,, numerical integration schemes, streamlines, streamline placement, geometric flow visualisation techniques, line integral convolution (LIC), texture- and feature-based flow visualisation.
The course introduces students to the mathematical techniques that scientists use to make statistically sound conclusions from their data. Topics include an introduction to Bayesian analysis, hypothesis testing, model fitting / selection, Monte Carlo Markov Chains (MCMC), Principal Component Analysis (PCA), among others. Students will get the chance to work with real data, and develop their coding abilities.
Machine Learning is the science of how we can build abstractions of the world from data. In this unit we will start with the fundamental underlying principles and philosophies that allows us to learn and then look at how we can formulate these using explicit models.
Machine learning is mathematical in nature and a good grasp of linear algebra and multi-variate calculus is required to fully digest the material.