CSC-20071 - Applied Deep Learning
Coordinator: Baidaa Al-Bander
Lecture Time: See Timetable...
Level: Level 5
Credits: 15
Study Hours: 150
School Office: 01782 733075

Programme/Approved Electives for 2022/23

None

Available as a Free Standing Elective

No

Co-requisites

None

Prerequisites

CSC-10058 Introduction to Data Science I
CSC-10060 Introduction to Data Science II
MAT-10055 Mathematical Techniques for Data Science

Barred Combinations

None

Description for 2022/23

This module will provide students with an introduction to deep learning, with a focus on understanding its capabilities and writing programs that use appropriate software libraries to apply deep learning to tasks such as text analysis, computer vision, image processing and pattern recognition.

Aims
The aim of the module is to provide students with an introduction to deep learning, with a focus on understanding its capabilities and writing programs that use a software library to apply deep learning to tasks such as pattern recognition and classification when applied to text processing, computer vision and image processing.

Intended Learning Outcomes

Identify and describe the capabilities and limitations of deep neural networks, including convolutional neural networks and long short term memory networks.: Develop software that uses appropriate libraries to create, train and evaluate deep neural networks.: Apply deep learning to tasks such as computer vision and textual sentiment analysis using techniques such as transfer learning.:

Study hours

15 hours lectures
20 hours practical classes
25 hours coursework preparation
90 hours private study

School Rules

CSC-20043 Computational and Artificial Intelligence 1

Description of Module Assessment

1: Coursework weighted 100%
Deep learning software development and recorded screencast
Students will be required to write a program or programs to create, train and evaluate a deep neural network for a given or chosen task; experiment with model architecture and parameter choices, commenting intelligently on their effect on results; and evaluate their results in the context of the capabilities and limitations of deep neural networks. Each student will submit their program code (equivalent to a 3000 word report) together with a "recorded screencast" or "live demo" (equivalent to a 2000 word report) of them using their software to train and evaluate their deep neural network; vary model architecture and parameters, commenting on results; and discuss their results in the context of the capabilities and limitations of deep neural networks.