Enhancing Efficiency of Foundational Models

The research will explore techniques such as model sparsity, knowledge distillation, and task-specific adaptation to reduce computational overhead without sacrificing performance. Additionally, we will investigate the use of more efficient optimization methods to speed up training and inference, making large-scale models more accessible and sustainable. The results will aim to provide scalable solutions that enable the continued progress of AI while reducing resource consumption.This project will be part the Data Science Research Group at the University of Queensland located in Brisbane.

This project aims to improve efficiency in deep learning models. This project will be part the Data Science Research Group, Information Technology and Electrical Engineering School, at the University of Queensland located in Brisbane, Australia.

The Data Science group researches and develops innovative and practical solutions for business, scientific and social applications in the realm of big data. The group encompasses a variety of research strengths including: Data and knowledge engineering, Information Retrieval, Computer Vision, and Complex and Intelligent Systems. You will join a world-leading research group currently composed of 13 academic staff members (including 6 full professors, two DECRA fellows and an 2 Future Fellows), 7 research fellows and over 40 PhD students. Members of the group have a successful track record of publishing in top conferences and journals such as ACM SIGIR, ACM CIKM, The Web Conference (WWW), SIGMOD, CVPR, ICCV, ICML, PAMI, JMLR, ICLR and various ACM and IEEE transactions.

The research environment available to the project is world-class. The University of Queensland (UQ) has a strong and internationally focused research culture. It is ranked in the top 1% of world universities in three widely publicized international University rankings.

This project is supported by the Research project scholarship.

Learn more about the Research project scholarship.

You must contact the principal supervisor for this project to discuss your interest. You should only complete the online application after you have reached agreement on supervision.

Always make sure you are approaching your potential supervisor in a professional way. We have provided some guidelines for you on how to contact a supervisor.

Your application will be assessed on a competitive basis.

We take into account your:

• previous academic record
• publication record
• honours and awards
• employment history

A working knowledge of solid programming and algorithmic skills, and knowledge of machine learning techniques, particularly deep learning would be of benefit to someone working on this project.

You will demonstrate academic achievement in the field/s of machine learning and computer vision and the potential for scholastic success.

A background or knowledge of linear algebra, and multimodal learning is highly desirable.

Before you apply

Before submitting an application you should:

1. check your eligibility for a Doctor of Philosophy (PhD)
2. prepare your documentation
3. contact Associate Professor Mahsa Baktashmotlagh (m.baktashmotlagh@uq.edu.au) to discuss your interest and suitability
4. submit your application by 10 April, 2025 10 April, 2025.

When you apply

You apply for this scholarship when you submit an application for your program. You don’t need to submit a separate scholarship application.

In your application ensure that under the ‘Scholarships and collaborative study’ section you select:

• ‘My higher degree is not collaborative’
• ‘I am applying for, or have been awarded a scholarship or sponsorship'
• ‘Other’, then ‘Research Project Scholarship’ and in the ‘Name of scholarship’ field enter EFFICIENCY-BAKTASHMOTLAGH.

Apply now