Module Information

Module Identifier
SEM6420
Module Title
Machine Learning for Intelligent Systems
Academic Year
2017/2018
Co-ordinator
Semester
Semester 2
Pre-Requisite
MEng Year 4
Other Staff

Course Delivery

Delivery Type Delivery length / details
Lecture 10 x 2 Hour Lectures
Practical 5 x 2 Hour Practicals
 

Assessment

Assessment Type Assessment length / details Proportion
Semester Assessment Written analysis of scientific paper(s)  (3000 words limit), followed by an oral presentation and discussion on the same.  40%
Semester Assessment Written assignment  contrasting the use of several methods discussed in the course, applied to data provided by the lecturers (4000 words).  60%
Supplementary Assessment Resubmission of failed/nonsubmitted components or others of equivalent value.  100%

Learning Outcomes

On successful completion of this module students should be able to:

Demonstrate competence with the machine learning methods and tools considered in this scheme.

Show proficiency in analysing data sets using the appropriate tools.

Demonstrate skills in designing, running and documenting experiments using machine learning.

Demonstrate capability to write and present a detailed analysis of an application of machine learning.

Brief description

This module will equip students with the main concepts in Machine Learning by engaging them in seminar-based discussions on scientific papers. It will then help the students build towards a term paper, which will describe their practical investigation of the issues involved in applying two machine learning methods to an appropriate data set that they will have found.

Content

The content will closely follow Alpaydin's book, with additional use of Mitchell's book. The lectures will introduce the ideas, and the students will be expected to read further from the book. This will be tested by getting them to do presentations on sections of the book not covered in class.

1. What is Machine Learning? (2 hrs)

Foundations and assumptions of ML.

2. Supervised Learning. (2 hrs)

Learning from labelled examples.

3. Bayesian Decision Theory. (3 hrs)

Probability and optimality in learning.

4. Parametric Methods. (2 hrs)

5. Dimensionality Reduction. (2 hrs)

Detecting unnecessary attributes and removing them to improve accuracy.

6. Clustering. (3 hrs)

K-means, hierarchical, consensus clustering techniques.

7. Nonparametric Methods. (3 hrs)

Learning without constructing a model (esp. kNN); transductive learning.

8. Hidden Markov models. (3 hrs)

Probabilistic, structural models from data.

9. Assessing and Comparing Classification Algorithms. (3 hrs)

10. Combining Multiple Learners. (3 hrs)

Obtained improved results by combining the predictions of multiple classifiers.

11. Reinforcement Learning. (2 hrs)

Learning sequences of actions with reward.

Additional material requested by students via questionnaire. (2 hrs)

Module Skills

Skills Type Skills details
Application of Number Inherent to subject
Communication Seminar
Improving own Learning and Performance Inherent to subject
Information Technology Inherent to subject
Personal Development and Career planning Encourages students to see roles in subject for career and personal development
Problem solving Inherent to subject
Research skills Essay
Subject Specific Skills Representation and Reasoning for Intelligent Systems
Team work

Notes

This module is at CQFW Level 7