MATH5191 is an Honours and Postgraduate Coursework Mathematics course. See the course overview below.

Units of Credit: 6

Prerequisites: N/A

Exclusion: MATH3191 Mathematical Optimization for Data Science (jointly taught with MATH5191)

Cycle of offering: Term 3 2022 & 2024 (even years)

Graduate attributes: The students in PG version (MATH 5191) are expected to be able to critically evaluate optimisation techniques, as well as modify and synthesise new methods for data science problems. 

More information:  The Course outline will be made available closer to the start of term - please visit this website: https://www.unsw.edu.au/course-outlines

Important additional information as of 2023

UNSW Plagiarism Policy

The University requires all students to be aware of its policy on plagiarism.

For courses convened by the School of Mathematics and Statistics no assistance using generative AI software is allowed unless specifically referred to in the individual assessment tasks.

If its use is detected in the no assistance case, it will be regarded as serious academic misconduct and subject to the standard penalties, which may include 00FL, suspension and exclusion.

The Online Handbook entry contains up-to-date timetabling information.

If you are currently enrolled in MATH5191 you can log into UNSW Moodle for this course.

Course aims

Introduce major mathematical ideas behind modern optimisation techniques used in data science, such as convex and nonconvex (continuous) optimisation problems, first-order methods, splitting and projection techniques, stochastic optimisation.

Discuss the considerations contributing to complexity analysis of optimisation problems and algorithms in the context of data science, such as the problem's size and structure, accuracy and efficiency requirements, advantages and limitations of different optimisation techniques, and different perspectives on convergence and (iteration) complexity.

Place optimisation techniques in the context of major data science applications such as the training of artificial neural networks and data classification, addressing the appropriate choice of numerical methods and their limitations.

Introduce the students to professional communication styles in the area of optimisation for data science, in particular mapping the ideas and terminology used in different fields. Help students develop effective communication strategies within the topic.

Course descriptions

The course covers theoretical foundations necessary for the in-depth understanding of modern optimisation methods for data science. The optimisation methods are presented in the context of relevant applications, such as the training of artificial neural networks and data classification.

The methods discussed in the course include (stochastic) gradient descent, projection and splitting techniques. The course prepares students for confident application of modern numerical methods to problems in data science and helps them build sufficient mastery of optimisation tools and techniques for designing and implementing tailored methods for solving new problems.