Overview

Indicative content includes:

• Basics of machine learning: Concepts of object, class, feature. Training and testing protocols. Error estimation. ROC curves. Supervised and unsupervised learning.
• Classification methods: basic classifiers and classifier ensembles.
• Feature selection.
• Clustering.
• Neural networks: standard architectures and deep learning.

Assessment Strategy

-threshold -Equivalent to 40%. The student is able to reason within the taught material to a satisfactory extent. They are familiar with the basic concepts of dataset, feature, class, class label, feature space, etc. They understand the basic models of classification and clustering and can apply off-the-shelf software to synthetic and real data.

-good -Equivalent to the range 60%-69%. The student demonstrates good understanding of the material. They are able to reproduce and apply basic algorithms for classification, clustering and feature selection. The student can apply given off-the-shelf algorithms to synthetic and real data sets.

-excellent -Equivalent to the range 70%+. The student demonstrates deep understanding of the material. They are able to reproduce and apply all the taught algorithms for classification, clustering, and feature selection. The student can choose appropriately and apply off-the-shelf algorithms to synthetic and real data sets.

Learning Outcomes

• Detail and apply clustering algorithms to data sets.

• Detail and apply various classification models.

• Explain and apply the basic notions and principles of machine learning.

• Explain dimensionality reduction, its approaches and methods.

• Summarise neural network models and their training procedures.