| Unit name | Applied Deep Learning |
|---|---|
| Unit code | COMSM0018 |
| Credit points | 10 |
| Level of study | M/7 |
| Teaching block(s) |
Teaching Block 1 (weeks 1 - 12) |
| Unit director | Professor. Damen |
| Open unit status | Not open |
| Pre-requisites |
COMS30007 Machine Learning |
| Co-requisites |
None |
| School/department | Department of Computer Science |
| Faculty | Faculty of Engineering |
The unit introduces the students to the latest deep architectures for learning linear and non-linear transformations of big data towards tasks such as classification and regression. The unit paves the path from understanding the fundamentals of convolutional and recurrent neural networks through to training and optimisation as well as evaluation of learnt outcomes. The unit’s approach is hands-on, focusing on the ‘how-to’ while covering the basic theoretical foundations.
On successful completion of this unit, you will: understand the opportunities and challenges that deep architectures, both convolutional and recurrent, bring to machine learning tasks such as classification and regression; appreciate the role of a variety of optimization approaches in parameter training for deep architectures; learn how to setup, train and evaluate deep architectures on public datasets; replicate published experiments and discuss the successes and failures of current architectures.
The teaching block will be split into three parts. Part one (3 weeks) will include lectures [2 hours per week]. Part two (4 weeks) will include a combination of lectures and hands-on labs [1 hour of lecturing and 2 hours of lab per week] and Part three (4 weeks) will include seminars for interim presentations (Weeks 9-10) and project work with the continuation of hands-on labs [2 hour of lab per week]. The ‘CS Explorers’ week will take place in Week 8 at the end of Part two, with continuing lab support. The material is front-loaded to ensure the fundamentals are covered before students attempt hands-on applied work.
The student will undertake a challenge of replicating a state-of-the-art performance on a publicly available dataset using one of the deep architectures discussed on the unit. This coursework will be assessed through an interim presentation (oral) or report (written) (40%) as well as a final report (60%). The unit will be able to assess the various learning outcomes including the ability to communicate results and reflect on the success and failure of the selected approach.
he unit will use a main textbook as well as a number of online resources. The textbook will be:
Goodfellow et al (2016). Deep Learning. MIT Press
Note that this unit is the first textbook to be made available on the topic. The textbook could change as more resources are made available
