This module covers the rationale, techniques and applications of Machine Learning. The aim of this module is to provide students with a comprehensive foundation and practical skills in machine learning including the underlying principles, the techniques used and applications of machine learning. Students will be required to use the techniques presented and to design and develop a machine learning system.
- Introduction to Machine Learning: Classification, regression, generalisation, noise; feature representation and selection; model selection; supervised and unsupervised learning; parametric and non-parametric methods; applications of machine learning;
- Information-Based Learning: Decision tree representation; basic decision tree learning algorithm ID3; choosing the best attribute, entropy and information gain; applications of decision trees
- Probability-Based Learning: Bayes rule, prior probability, Naïve Bayes classifier
- Similarity-Based Learning: k-Nearest Neighbour method; locally weighted regression; Case- Based Reasoning; lazy versus eager learning; applications of instance-based learning;
- Error-Based Learning: Multiple Variable Linear Regression; Gradient Descent; Multinomial Logistic Regression; applications of MLR, Support Vector Machines, Neural Networks
- Evaluation: Performance Measures, Experimental Methodology, Module Selection, Data Splitting Techniques
- Dimensionality Reduction: Feature Selection, Principle Components Analysis;
- Combining Multiple Learners: Ensemble learning, diversity in ensembles; base classifier generation methods, bagging, boosting; ensemble aggregation techniques;
- Unsupervised Learning: Cluster analysis; applications of clustering; partitioning algorithms; k-means method; hierarchical method; cluster validity approaches.
- Issues in Machine Learning: “curse of dimensionality”; concept drift; directions in machine learning research; overfitting
The module is designed to be delivered within a blended learning model, employing mixed modes (online and face to face) of learning, teaching and assessment.
TU059 will be delivered primarily in a face-to-face mode while TU256 and TU060 will be delivered in a blended mode.
This module will use lectures and supervised lab-based sessions including face-to-face and/or online delivery and independent learning through assignments.
The practical element of the module will be supported through lab sessions where students will be given examples of the techniques and will be expected to apply these techniques to sample problems
Students will be encouraged to be pro-active in their approach to learning through the use of practical examples of the application of the algorithms. Students will be expected to develop independence in, and take responsibility for their own learning.
| Module Content & Assessment | |
|---|---|
| Assessment Breakdown | % |
| Formal Examination | 50 |
| Other Assessment(s) | 50 |