Pattern Recognition and Analysis (COMP3710)

Course overview

Understanding patterns in our environment is an important cognitive ability. The development of recognition and automated algorithms that are able to process copious amounts data without (or with limited) human intervention is critical in replicating this ability in machines. This course will cover the fundamentals of creating computational algorithms and models that are able to recognise and/or analyse patterns within data of various forms. Topics and algorithms will include fractal geometry, classification methods such as random forests, recognition approaches using deep learning and models of the human vision system. Python, the open-source paradigm and state-of-the-art packages like Tensorflow, JAX and PyTorch will be used as a mechanism for students to study patterns in nature and solve problems from various real-worlds data sources, such images, social media and biomedical signals.

The course will be taught via three main modules:

1. Patterns in nature – how symmetry and self-similarity leads to the patterns we observe in nature, such as fractals and the known particles of the universe.
2. Traditional pattern recognition – how patterns are quantified, measured and identified using features, similarity, information theory, principles of transform domains and decision trees.
3. Deep Learning – how the idea of receptive fields, convolutions and attention mechanisms are used to learn appropriate multi-scale, self-adaptive features via neural networks.

In general, students will have opportunities to implement and create pattern recognition and analysis solutions using the algorithms discussed on actual research data from various fields. This will allow them to specialise into areas of image/signal analysisﾠand data science. Guest lecturers and leaders in some of these areas will present some of their work to highlight the societal impact of these algorithms.

Course requirements

Assumed background

This course assumes that students are confident and knowledgeable in Python programming and linear algebra. You will be required to learn and keep up with implementing algorithms in Tensorflow/PyTorch frameworks. A (non-assessed) basic Python practical (without Tensorflow/PyTorch) will be provided in the first weekﾠas a refresher for students.

Students will be required to use version control to manage your code, so two short courses on Git are required to be completed during the course preferably before the report assessment.

Masters Students: Masters students are allowed to enrol in a small number of undergraduate courses in a recent rule change, so you may be able to take COMP3710. Students should consult with an advisor to make sure it will fit within their program(s).

Prerequisites

You'll need to complete the following courses before enrolling in this one:

(MATH1051 or MATH1071) and (CSSE1001 or ENGG1001)

Recommended prerequisites

We recommend completing the following courses before enrolling in this one:

(MATH1052 or MATH1072) and MATH2302 and COMP3506

Course contact

Course staff

Timetable

The timetable for this course is available on the UQ Public Timetable.

Additional timetable information

Lectures will also be streamed via ZOOM, recorded and held live on campus simultaneously. Course notes for some modules of the course will also be provided.

Aims and outcomes

The course will provide students with practical experiences in:

• solving pattern recognition problems, especially via deepﾠneural networks
• a deep learning framework, such as Tensorflow, JAX or PyTorch
• algorithmic design incorporating software engineering principles,
• Open source development through creating and/orﾠcontributing to a course run open source project, such as PatternFlow – a pattern recognition and image processing library for deep learning
• mathematical frameworks involved in understanding patterns and principles required for all of the above, such as group theory, fractal geometry and tensor calculus

ﾠ

Learning outcomes

After successfully completing this course you should be able to:

Assessment

Assessment summary

A hurdle is an assessment requirement that must be satisfied in order to receive a specific grade for the course. Check the assessment details for more information about hurdle requirements.

Assessment details

Fractals

• In-person

Mode

Activity/ Performance, Oral

Category

Computer Code, Practical/ Demonstration

Weight

15%

Due date

5/08/2024 - 19/08/2024

Assessment during scheduled lab session

Other conditions

Student specific.

See the conditions definitions

Learning outcomes

L01, L02

Task description

Submission guidelines

Deferral or extension

You cannot defer or apply for an extension for this assessment.

This course uses Demos that are allocated 3–5-week blocks of schedule practical sessions in order to complete them. Marking is to be done before the end of each block in the scheduled practical times where possible. You cannot defer these, and no extensions will be available because these involve an oral explanation/demo in the scheduled lab session. If there are exceptional circumstances an exemption may be approved and may involve submitting/discussing your work as it stands. Exemptions must be requested as an extension with a note specifying exemption via my.UQ.

Late submission

You will receive a mark of 0 if this assessment is submitted late.

Assessment items received after the deadline will be subject to a late penalty of 100% after grace period of 1-hour.

Pattern Recognition

• In-person

Mode

Activity/ Performance, Oral

Category

Computer Code, Practical/ Demonstration

Weight

25%

Due date

2/09/2024 - 16/09/2024

Assessment during scheduled lab session

Other conditions

Student specific.

See the conditions definitions

Learning outcomes

L02, L03, L04, L05

Task description

Submission guidelines

Deferral or extension

You cannot defer or apply for an extension for this assessment.

This course uses Demos that are allocated 3–5-week blocks of schedule practical sessions in order to complete them. Marking is to be done before the end of each block in the scheduled practical times where possible. You cannot defer these, and no extensions will be available because these involve an oral explanation/demo in the scheduled lab session. If there are exceptional circumstances an exemption may be approved and may involve submitting/discussing your work as it stands. Exemptions must be requested as an extension with a note specifying exemption via my.UQ.

Late submission

You will receive a mark of 0 if this assessment is submitted late.

Assessment items received after the deadline will be subject to a late penalty of 100% after grace period of 1-hour.

Pattern Analysis Project

• Hurdle
• Online

Mode

Activity/ Performance, Written

Category

Computer Code, Paper/ Report/ Annotation, Project

Weight

30%

Due date

25/10/2024 4:00 pm

Assessment during scheduled lab session + Submission items

Other conditions

Student specific.

See the conditions definitions

Learning outcomes

L02, L03, L06, L07, L08

Task description

Hurdle requirements

Submission guidelines

This will be a Turn-it-in submission and a pull request to the relevant open source project.

Deferral or extension

You may be able to apply for an extension.

The maximum extension allowed is 7 days. Extensions are given in multiples of 24 hours.

This course uses allocated 3–5-week blocks of schedule practical sessions in order to complete the project. Only one extension will be available because these involve a turn-it-in and pull request submissions if there are exceptional circumstances and must be requested as an extension via my.UQ.

Late submission

You will receive a mark of 0 if this assessment is submitted late.

Assessment items received after the deadline will be subject to a late penalty of 100% after grace period of 1-hour.

Final Exam

• Hurdle
• Identity Verified
• In-person

Mode

Written

Category

Examination

Weight

30%

Due date

End of Semester Exam Period

2/11/2024 - 16/11/2024

Learning outcomes

L01, L02, L03, L04, L05, L06, L07, L08

Task description

Hurdle requirements

Exam details

Submission guidelines

Deferral or extension

You may be able to defer this exam.

Course grading

Full criteria for each grade is available in the Assessment Procedure.

Additional course grading information

Notes that 'passing' assessments is defined as obtaining 50% or more on that assessment. Marks will be rounded up for computing the grade hurdles and the final grade.

Supplementary assessment

Supplementary assessment is available for this course.

Additional assessment information

Use of AI Tools

Artificial Intelligence (AI) and Machine Translation (MT) are emerging tools that may support students in completing this assessment task. Students may appropriately use AI and/or MT in completing this assessment task. Students must clearly reference any use of AI or MT in each instance. A failure to reference generative AI or MT use may constitute student misconduct under the Student Code of Conduct

Having Trouble?

If you are having difficulties with any aspect of the course material you should seek help. Speak to the course teaching staff.

If external circumstances are affecting your ability to work on the course, you should seek help as soon as possible. The University and UQ Union have organisations and staff who are able to help, for example, UQ Student Services are able to help with study and exam skills, tertiary learning skills, writing skills, financial assistance, personal issues, and disability services (among other things).

Complaints and criticisms should be directed in the first instance to the course coordinator. If you are not satisfied with the outcome, you may bring the matter to the attention of the School of EECS Director of Teaching and Learning.

Learning resources

You'll need the following resources to successfully complete the course. We've indicated below if you need a personal copy of the reading materials or your own item.

Library resources

Find the required and recommended resources for this course on the UQ Library website.

Additional learning resources information

Dr Chandra will provide additional lecture notes on Blackboard for all modules of the course.

Watch the Pattern Analysis YouTubeﾠPlaylist of the covered topics created for this course.

Learning activities

The learning activities for this course are outlined below. Learn more about the learning outcomes that apply to this course.

Policies and procedures

University policies and procedures apply to all aspects of student life. As a UQ student, you must comply with University-wide and program-specific requirements, including the:

• Student Code of Conduct Policy
• Student Integrity and Misconduct Policy and Procedure
• Assessment Procedure
• Examinations Procedure
• Reasonable Adjustments - Students Policy and Procedure

Learn more about UQ policies on my.UQ and the Policy and Procedure Library.

School guidelines

Your school has additional guidelines you'll need to follow for this course:

• EECS School Guidelines for Students