CQUniversity Unit Profile
COIT29225 Neural Networks and Deep Learning
Neural Networks and Deep Learning
All details in this unit profile for COIT29225 have been officially approved by CQUniversity and represent a learning partnership between the University and you (our student).
The information will not be changed unless absolutely necessary and any change will be clearly indicated by an approved correction included in the profile.
General Information

Overview

Artificial Intelligence (AI) is becoming an important part of software development. Neural networks and Deep Learning are the main contributors to the recent advances in applications of Artificial Intelligence. Deep Learning enables computers to learn complicated concepts by building them out of a hierarchy of simpler ones. Deep Learning techniques have been successfully applied to a broad field of applications such as computer vision, image and video recognition, natural language processing, and medical diagnosis. This unit introduces you to the fundamentals of Deep Learning and how it can solve problems in many areas. In this unit, you will learn the architecture of neural networks and algorithms, including the latest Deep Learning techniques. You will learn to develop conventional neural networks such as multilayer perceptrons, and convolutional neural networks. You will use software to train and deploy neural networks. You will also identify practical applications of Deep learning by exploring recent case studies.

Details

Career Level: Postgraduate
Unit Level: Level 9
Credit Points: 6
Student Contribution Band: 8
Fraction of Full-Time Student Load: 0.125

Pre-requisites or Co-requisites

Pre-requisite: COIT20277 Introduction to Artificial Intelligence

Important note: Students enrolled in a subsequent unit who failed their pre-requisite unit, should drop the subsequent unit before the census date or within 10 working days of Fail grade notification. Students who do not drop the unit in this timeframe cannot later drop the unit without academic and financial liability. See details in the Assessment Policy and Procedure (Higher Education Coursework).

Offerings For Term 2 - 2020

Brisbane
Melbourne
Online
Sydney

Attendance Requirements

All on-campus students are expected to attend scheduled classes – in some units, these classes are identified as a mandatory (pass/fail) component and attendance is compulsory. International students, on a student visa, must maintain a full time study load and meet both attendance and academic progress requirements in each study period (satisfactory attendance for International students is defined as maintaining at least an 80% attendance record).

Class and Assessment Overview

Recommended Student Time Commitment

Each 6-credit Postgraduate unit at CQUniversity requires an overall time commitment of an average of 12.5 hours of study per week, making a total of 150 hours for the unit.

Class Timetable

Bundaberg, Cairns, Emerald, Gladstone, Mackay, Rockhampton, Townsville
Adelaide, Brisbane, Melbourne, Perth, Sydney

Assessment Overview

1. Practical Assessment
Weighting: 30%
2. Practical Assessment
Weighting: 35%
3. Project (applied)
Weighting: 35%

Assessment Grading

This is a graded unit: your overall grade will be calculated from the marks or grades for each assessment task, based on the relative weightings shown in the table above. You must obtain an overall mark for the unit of at least 50%, or an overall grade of ‘pass’ in order to pass the unit. If any ‘pass/fail’ tasks are shown in the table above they must also be completed successfully (‘pass’ grade). You must also meet any minimum mark requirements specified for a particular assessment task, as detailed in the ‘assessment task’ section (note that in some instances, the minimum mark for a task may be greater than 50%). Consult the University’s Grades and Results Policy for more details of interim results and final grades.

Unit Learning Outcomes
On successful completion of this unit, you will be able to:
  1. Formulate a neural network and deep learning problem applying the concepts and theory of classical and deep learning techniques
  2. Design deep learning solutions to problems in pattern recognition and image analysis
  3. Build a software application implementing neural networks using a high level programming language
  4. Evaluate the performance of the deep learning techniques used in the software application
  5. Investigate the application of intelligent systems in socially innovative applications.

The Skills Framework for the Information Age (SFIA) standard covers the skills and competencies related to information and communication technologies. SFIA defines levels of responsibility and skills. SFIA is adopted by organisations, governments and individuals in many countries. SFIA is increasingly being used when developing job descriptions and role profiles. SFIA can be used by individuals for creating personal skills profile. The Australian Computer Society (ACS) recognises the SFIA and provides MySFIA for ACS members to build a skills profile.

This unit contributes to the following workplace skills as defined by SFIA 7 (the SFIA code is included):

Software design (SWDN)

Programming/software development (PROG)

Testing (TEST)

Application Support (ASUP).

Alignment of Learning Outcomes, Assessment and Graduate Attributes
N/A Level
Introductory Level
Intermediate Level
Graduate Level
Professional Level
Advanced Level

Alignment of Assessment Tasks to Learning Outcomes

Assessment Tasks Learning Outcomes
1 2 3 4 5
1 - Practical Assessment - 30%
2 - Practical Assessment - 35%
3 - Project (applied) - 35%

Alignment of Graduate Attributes to Learning Outcomes

Graduate Attributes Learning Outcomes
1 2 3 4 5
1 - Knowledge
2 - Communication
3 - Cognitive, technical and creative skills
4 - Research
5 - Self-management
6 - Ethical and Professional Responsibility
7 - Leadership
8 - Aboriginal and Torres Strait Islander Cultures

Alignment of Assessment Tasks to Graduate Attributes

Assessment Tasks Graduate Attributes
1 2 3 4 5 6 7 8
1 - Practical Assessment - 30%
2 - Practical Assessment - 35%
3 - Project (applied) - 35%
Textbooks and Resources

Textbooks

Prescribed

Neural Networks and Deep Learning - A Textbook

Edition: 1st (2018)
Authors: Charu C. Aggarwal
Springer
Gewerbestrasse Gewerbestrasse , Cham , Switzerland
ISBN: 978-3-319-94462-3, 978-3-319-94463-0(eBook)
Binding: Paperback

Additional Textbook Information

If you prefer to study with a paper copy, they are available at the CQUni Bookshop here: http://bookshop.cqu.edu.au (search on the Unit code). eBooks are available at the publisher's website.

IT Resources

You will need access to the following IT resources:
  • CQUniversity Student Email
  • Internet
  • Unit Website (Moodle)
  • Python 3.7 or higher
  • TensorFlow and Keras
  • Anaconda3 2019.10 or 2020.02
Referencing Style

No referencing style set.

Teaching Contacts
Michael Li Unit Coordinator
m.li@cqu.edu.au
Schedule
Week 1 Begin Date: 13 Jul 2020

Module/Topic

Fundamentals of Python Programming

Chapter

Lecture Notes

Events and Submissions/Topic

Week 2 Begin Date: 20 Jul 2020

Module/Topic

Numpy, Scipy, Matplotlib, Pandas and Scikits

Chapter

Lecture Notes

Events and Submissions/Topic

Week 3 Begin Date: 27 Jul 2020

Module/Topic

An Introduction to Neural Networks

Chapter

Textbook Chapter 1

Events and Submissions/Topic

Week 4 Begin Date: 03 Aug 2020

Module/Topic

Machine Learning with Shallow Neural Networks

Chapter

Textbook Chapter 2

Events and Submissions/Topic

Week 5 Begin Date: 10 Aug 2020

Module/Topic

Deep Neural Networks and Learning Algorithm (1)

Chapter

Textbook Chapter 3

Events and Submissions/Topic

Vacation Week Begin Date: 17 Aug 2020

Module/Topic

Chapter

Events and Submissions/Topic

Week 6 Begin Date: 24 Aug 2020

Module/Topic

Deep Neural Networks and Learning Algorithm (2)

Chapter

Textbook Chapters 3&4

Events and Submissions/Topic

Assignment 1 Due: Week 6 Friday (28 Aug 2020) 11:45 pm AEST
Week 7 Begin Date: 31 Aug 2020

Module/Topic

Radial Basis Function Networks and PCA

Chapter

Textbook Chapter 5 & Lecture Notes

Events and Submissions/Topic

Week 8 Begin Date: 07 Sep 2020

Module/Topic

Convolutional Neural Networks (1)

Chapter

Textbook Chapter 8 - Part 1

Events and Submissions/Topic

Week 9 Begin Date: 14 Sep 2020

Module/Topic

Convolutional Neural Networks (2)

Chapter

Textbook Chapter 8 - Part 2

Events and Submissions/Topic

Week 10 Begin Date: 21 Sep 2020

Module/Topic

Convolutional Neural Networks (3)

Chapter

Textbook Chapter 8 - Part 3 & Lecture Notes

Events and Submissions/Topic

Assignment 2 Due: Week 10 Friday (25 Sept 2020) 11:45 pm AEST
Week 11 Begin Date: 28 Sep 2020

Module/Topic

Restricted Boltzmann Machines

Chapter

Textbook Chapter 6

Events and Submissions/Topic

Week 12 Begin Date: 05 Oct 2020

Module/Topic

Recurrent Neural Networks

Chapter

Textbook Chapter 7

Events and Submissions/Topic

Review/Exam Week Begin Date: 12 Oct 2020

Module/Topic

Chapter

Events and Submissions/Topic

Project Due: Review/Exam Week Monday (12 Oct 2020) 11:45 pm AEST
Exam Week Begin Date: 19 Oct 2020

Module/Topic

Chapter

Events and Submissions/Topic

Term Specific Information

Unit Coordinator: Dr. Michael Li 

email: m.li@cqu.edu.au

contact phone: (07)49306337

Assessment Tasks

1 Practical Assessment

Assessment Title
Assignment 1

Task Description

In this assignment, you will demonstrate your ability to analyze the given scenario, and formulate a problem that can be resolved by neural network techniques. You will identify the input, output, and the required parameters of the feedforward neural network and apply the appropriate learning algorithm. You will design and implement the software solutions that need to use python programming skills, including the use of the built-in python advanced libraries (such as NumPy, SciPy, Matplotlib and Scikits, etc.). The assessment task is to design, code, debug, and test the software applications using the topics in Weeks 1- 5. The assessment task may include two sub-tasks. Further details are in the Assignment 1 specification document available from the Unit website.


Assessment Due Date

Week 6 Friday (28 Aug 2020) 11:45 pm AEST


Return Date to Students

Week 8 Friday (11 Sept 2020)


Weighting
30%

Assessment Criteria

1. Clear analysis of the given scenario and user requirements

2. Correct use of relevant python libraries and good programming practice

3. Building the correct architecture of the proposed neural network

4. Correct use and implementation of the learning algorithm

5. A report documenting testing, and specific aspect of

    feedforward neural networks as required.


Submission
Online

Submission Instructions
Submit only one zip file (.zip) containing the source code files and the Word document file.

Learning Outcomes Assessed
  • Formulate a neural network and deep learning problem applying the concepts and theory of classical and deep learning techniques
  • Design deep learning solutions to problems in pattern recognition and image analysis


Graduate Attributes
  • Knowledge
  • Communication
  • Cognitive, technical and creative skills

2 Practical Assessment

Assessment Title
Assignment 2

Task Description

Assignment 2 contains two computational tasks that can be resolved by applying neural network techniques. The first task is a typical regression problem with the noisy data sample. You are required to apply the radial basis function (RBF) network or its variant – extreme learning machines (ELM) to implement a software solution. The second task is a classification problem, which may connect to a standard benchmark dataset. You are required to build a deep neural network to provide a solution. In both tasks, you need to design, code, debug, and test the software solutions using the topics in Weeks 4 - 7. Further details are in the Assignment 2 specification document available from the Unit website.


Assessment Due Date

Week 10 Friday (25 Sept 2020) 11:45 pm AEST


Return Date to Students

Week 12 Friday (9 Oct 2020)


Weighting
35%

Assessment Criteria

1. Data format represented/generated correctly

2. Correct modelling of problem solutions following neural network concepts

3. Reasonable results with MSE and classification accuracy

4. Effective use of good programming practice in the code development

5. Rigorous testing to evaluate the correctness of the model and algorithm

6. A report discussing the design, implementation, and test of proposed networks


Submission
Online

Submission Instructions
Submit only one zip file (.zip) containing the source code files and the Word document file.

Learning Outcomes Assessed
  • Formulate a neural network and deep learning problem applying the concepts and theory of classical and deep learning techniques
  • Design deep learning solutions to problems in pattern recognition and image analysis
  • Build a software application implementing neural networks using a high level programming language
  • Investigate the application of intelligent systems in socially innovative applications.


Graduate Attributes
  • Knowledge
  • Communication
  • Cognitive, technical and creative skills
  • Research
  • Ethical and Professional Responsibility
  • Leadership

3 Project (applied)

Assessment Title
Project

Task Description

In this project, you will demonstrate your ability to apply the convolutional neural networks (CNN) and deep learning algorithm to build an image classifier. The performance and accuracy of your CNN network solution should be analyzed and discussed properly. You also need to compare the advantages and disadvantages between the CNN technique and the traditional feedforward neural network method. The assessment task is to design, code, debug, and test a software application using the topics in Weeks 8 - 10. Further details are in the Project specification document available from the Unit website.


Assessment Due Date

Review/Exam Week Monday (12 Oct 2020) 11:45 pm AEST


Return Date to Students

Exam Week Monday (19 Oct 2020)


Weighting
35%

Assessment Criteria

1. Clear analysis of the required classification problem

2. Correct design of the proposed CNN and parameter configurations

3. Correct applying deep learning principle

4. Effective source code development applying correct algorithmic steps

5. Writing a report to discuss the advantages & disadvantages,

    and potential practical applications of CNN


Submission
Online

Submission Instructions
Submit only one zip file (.zip) containing the source code files and the Word document file.

Learning Outcomes Assessed
  • Build a software application implementing neural networks using a high level programming language
  • Evaluate the performance of the deep learning techniques used in the software application
  • Investigate the application of intelligent systems in socially innovative applications.


Graduate Attributes
  • Knowledge
  • Cognitive, technical and creative skills
  • Research
  • Self-management
  • Ethical and Professional Responsibility
  • Leadership

Academic Integrity Statement

As a CQUniversity student you are expected to act honestly in all aspects of your academic work.

Any assessable work undertaken or submitted for review or assessment must be your own work. Assessable work is any type of work you do to meet the assessment requirements in the unit, including draft work submitted for review and feedback and final work to be assessed.

When you use the ideas, words or data of others in your assessment, you must thoroughly and clearly acknowledge the source of this information by using the correct referencing style for your unit. Using others’ work without proper acknowledgement may be considered a form of intellectual dishonesty.

Participating honestly, respectfully, responsibly, and fairly in your university study ensures the CQUniversity qualification you earn will be valued as a true indication of your individual academic achievement and will continue to receive the respect and recognition it deserves.

As a student, you are responsible for reading and following CQUniversity’s policies, including the Student Academic Integrity Policy and Procedure. This policy sets out CQUniversity’s expectations of you to act with integrity, examples of academic integrity breaches to avoid, the processes used to address alleged breaches of academic integrity, and potential penalties.

What is a breach of academic integrity?

A breach of academic integrity includes but is not limited to plagiarism, self-plagiarism, collusion, cheating, contract cheating, and academic misconduct. The Student Academic Integrity Policy and Procedure defines what these terms mean and gives examples.

Why is academic integrity important?

A breach of academic integrity may result in one or more penalties, including suspension or even expulsion from the University. It can also have negative implications for student visas and future enrolment at CQUniversity or elsewhere. Students who engage in contract cheating also risk being blackmailed by contract cheating services.

Where can I get assistance?

For academic advice and guidance, the Academic Learning Centre (ALC) can support you in becoming confident in completing assessments with integrity and of high standard.

What can you do to act with integrity?