COMP6208 Advanced Machine Learning
Module Overview
- To introduce key concepts in pattern recognition and machine learning; including specific algorithms for classification, regression, clustering and probabilistic modeling. - To give a broad view of the general issues arising in the application of algorithms to analysing data, common terms used, and common errors made if applied incorrectly. - To demonstrate a toolbox of techniques that can be immediately applied to real world problems, or used as a basis for future research into the topic.
Aims and Objectives
Module Aims
To provide an overview of advanced machine learning
Learning Outcomes
Knowledge and Understanding
Having successfully completed this module, you will be able to demonstrate knowledge and understanding of:
- Key concepts, tools and approaches for pattern recognition on complex data sets
- Kernel methods for handling high dimensional and non-linear patterns
- State-of-the-art algorithms such as Support Vector Machines and Bayesian networks
- Theoretical concepts and the motivations behind different learning frameworks
Subject Specific Practical Skills
Having successfully completed this module you will be able to:
- Be able to solve real-world machine learning tasks: from data to inference
Subject Specific Intellectual and Research Skills
Having successfully completed this module you will be able to:
- Conceptually understand the role of pattern analysis and probabilistic modeling, together with the mathematical techniques this requires
Syllabus
Key concepts - Supervised/Unsupervised Learning - Loss functions and generalization - Probability Theory - Parametric vs Non-parametric methods - Elements of Computational Learning Theory Kernel Methods for non-linear data - Support Vector Machines - Kernel Ridge Regression - Structure Kernels - Kernel PCA - Latent Semantic Analysis Bayesian methods for using prior knowledge and data - Bayesian inference - Bayesian Belief Networks and Graphical models - Probabilistic Latent Semantic Analysis - The Expectation-Maximisation (EM) algorithm - Gaussian Processes Ensemble Learning - Bagging - Boosting - Random Forest Dimensionality Reduction - CCA, LDA, ICA, NMF - Canonical Variates - Feature Selection vs Feature Extraction - Filter Methods - Sub-space approaches - Embedded methods Low-Rank approaches - Recommender Systems Application areas - Security - Business - Scientific
Learning and Teaching
| Type | Hours |
|---|---|
| Independent Study | 104 |
| Teaching | 46 |
| Total study time | 150 |
Resources & Reading list
Christopher M. Bishop (2006). Pattern Recognition and Machine Learning.
Information Theory, Inference, and Learning Algorithms.
Reasoning and Machine Learning.
John Shawe-Taylor and Nello Cristianini (2004). Kernel Methods for Pattern Analysis.
Assessment
Summative
| Method | Percentage contribution |
|---|---|
| Exam (2 hours) | 66.667% |
| Report | 33.333% |
Referral
| Method | Percentage contribution |
|---|---|
| Exam (2 hours) | 100% |
Repeat Information
Repeat type: Internal & External
Linked modules
Pre-requisites: COMP3206 OR COMP6229