Details on the 2024 AIML Summer Research Scholarship program coming soon.
Are you a current 成人大片 undergraduate student interested in machine learning research?
The Australian Institute for Machine Learning (AIML) is offering several paid summer research scholarships for students seeking to explore what postgraduate machine learning research has to offer.
Scholarships are available to continuing undergraduate students to undertake a six-week research project with AIML researchers.
What's in it for you?
Real research experience 鈥Based at the AIML building at Lot Fourteen, you鈥檒l be working with leading AIML researchers on cutting-edge machine learning projects. Use this hands-on experience to get a head start planning your postgraduate research career before completing your undergraduate studies.
Financial support 鈥 Each scholarship offers a one-off payment of $1,200. In addition, you鈥檒l get up to $800 to put towards equipment costs.
Research sprint 鈥 Focus on a six-week intensive research project during the summer semester.
Project 1: Exploring X-ray velocimetry (XV) imaging and machine learning for the detection and monitoring of cystic fibrosis disease in preclinical data
Cystic fibrosis is a life-limiting genetic condition wherein abnormally thick and sticky mucus is produced, causing complications that predominantly affect the respiratory, digestive and reproductive systems. Lung function is currently measured via a number of techniques, including spirometry and chest X-Ray imaging.
While valuable, each test provides partial information making it challenging to obtain a comprehensive assessment of the patient's condition. Our team is currently investigating the use of chest X-ray velocimetry (XV) imaging, a novel approach providing 4D (3D + time) functional information across the whole lung, and throughout the breath, for monitoring of cystic fibrosis disease and treatment.
Aim:
The proposed project will use machine learning techniques to explore whether the differentiation between healthy and cystic fibrosis lungs is possible on preclinical data. To achieve this, the successful candidate will:
perform literature review on the latest developments in cystic fibrosis imaging
curate and pre-process a preclinical XV dataset
develop algorithms that take XV images and classify healthy vs cystic fibrosis lungs
create qualitative and quantitative assessment tools for the algorithm results
report the findings and proceed to publishing the methodology and results
Advantages:
This project is a collaborative effort between the Australian Institute for Machine Learning (AIML), the Women鈥檚 and Children鈥檚 Hospital (WCH), and the Cystic Fibrosis Airway Research Group (CFARG). The project provides a great opportunity to work on cutting edge research on cystic fibrosis, developing and testing state-of-the-art AI algorithms for monitoring this life-limiting genetic condition. There is a strong clinical and translational focus and the successful candidate will get the opportunity to join an established team of scientists/clinicians working at the confluence of healthcare and technology.
Profile:
The successful candidate will need to be passionate about applied artificial intelligence solutions in healthcare applications. Experience in Python, machine learning and deep learning tools (both for data analysis and algorithm development), as well as interest in developing strong scientific writing skills is highly desirable.
This project aims to develop effective machine learning algorithms to predict CO2 capture and recycle materials. To mitigate the effects of climate change, one potential solution is to store and transfer the CO2 that is released into the environment.
Quantum chemistry simulations have been employed to guide the development of catalysts that are both cost-effective and efficient in converting CO2 into valuable fuels. Nonetheless, these simulations pose computational demands and encounter limitations regarding the quantity of samples, length scale, and time scale they can handle.
In this project, we seek to leverage machine learning (graph neural network and active learning) to accelerate these simulations and predict novel CO2 capture materials.
Skills and knowledge requirements:
The project necessitates a proficiency in programming languages like C++ and Python and deep learning frameworks like Pytorch.
Project 3: Semi-supervised active learning for object detection and 3D reconstruction
Supervisor:
Description:
Object detection and 3D reconstruction are fundamental tasks in computer vision, involving the identification of objects' presence and location in images or videos. However, these tasks become challenging in complex and dynamic environments.
The primary objective of this project is to develop a pseudo-annotation mining strategy to discover stable predictions-based approaches for object detection and 3D reconstruction under complex environments (e.g. occlusion, changing illumination), specifically tailored to agricultural related datasets. We aim to combine active learning and semi-supervised learning with large language models (LLMs), to enhance semantic understanding of an object and its 3D information. Our ultimate goal is to improve the accuracy of object detection and 3D shape reconstruction, thereby enabling the development of efficient computer vision systems with a wide range of applications.
Generative adversarial networks (GANs) are popular for generating synthetic, realistic and high-quality images such as faces. GANs are also used in many computer vision tasks such as data augmentation and drug discovery. We can鈥檛 help wondering how can these GANs generate a real and distinctive image for a random latent vector? This project aims to address this question by proposing new evaluation metric for explaining GAN models.
Project 6: Embodied AI - Imitation learning for robot navigation
Supervisors:
, ,
Description:
Robot navigation refers to the ability of a robot to autonomously navigate in a known or unknown environment to complete a task, e.g., delivering pizza or conducting an inspection.
Existing methods based on reinforcement learning or classical sense-plan-act pipelines are either limited to simulated environments or require fine-tuning to specific environment conditions (e.g., indoor or outdoor only).
In this project, the students will get hands-on experience operating a robot dog (Unitree Go1) to collect real-world demonstrations and use it to learn an imitation behaviour for autonomous navigation. Other sources for imitation, for example, through online videos of human/animal navigation would also be considered and annotated for this task. The project will focus on the varied potential sources for imitation and varied methods of converting that information to make a robot move under different circumstances, e.g., smooth trajectories around dynamic objects.
Eligibility
Applicants must be students who are:
currently enrolled in a continuing undergraduate program at the 成人大片
Australian citizens, permanent residents of Australia, New Zealand Citizens, holders of a permanent humanitarian visa and international students who hold current appropriate visas
Scholarship positions are strictly limited, and applications will be assessed on merit. Applicants should read the scholarship rules before applying.
Scholarship rules
Be sure to read the following scholarship rules while considering your application:
The Scholarships shall be called the 鈥淎IML Summer Research Scholarships鈥.
To be eligible for the Scholarship, candidates must:-
be an Australian citizen, permanent resident of Australia, New Zealand Citizen, holder of a Permanent Humanitarian Visa or an international student who holds current appropriate visas permitting study in Australia beyond February 2024;
be a current undergraduate or masters student by coursework of the 成人大片;
have completed at least two years (48 units) of study of their program, at a standard acceptable for admission to an Honours program (In exceptional circumstances, students who are completing their first year of the program may be considered, but only after consideration of the students who have completed second year);
be undertaking a six-week research project at the Australian Institute for Machine Learning over the 2023-24 summer vacation;
have an appropriate research project available within the student鈥檚 Faculty/School. (It is the responsibility of students to contact a supervisor to ensure that an appropriate research project is available).
Students who are not studying at the 成人大片 can apply for the Adelaide Summer Research Scholarship but must be enrolled to study at another University.
Students who are in receipt of, or expect to receive, any other summer or vacation scholarships over the 2023-2024 summer vacation (e.g. National Heart Foundation Summer Scholarship, Cancer Council Vacation Scholarships) are not eligible for the AIML Summer Research Scholarship unless the other scholarship is for a separate project.
The scholarship duration is six weeks during the 2023-2024 summer semester.
Up to 6 Scholarships shall be awarded on the recommendation of the relevant AIML supervisors, to students of high academic merit, and the availability of a relevant research project within AIML, which would be beneficial and of interest to each student selected. As students will need to be matched to suitable research projects, the scholarships may not be awarded strictly based on academic merit.
The scholarships are intended to provide experience in a research project conducted by one or more academic staff and are not intended to support work towards an Honours degree, or for any other course requirements.
Each scholarship will provide a living allowance of $1,200 as a one-off payment paid direct to the student by January 31st Tax payments are not deducted from this payment. In addition to the scholarship payment to the student, $800 will be available for equipment for the project. Recipients of the scholarship are responsible for seeking advice from the Australian Taxation Office regarding the status of the scholarship with regard to taxation.
The Scholarship must be taken up in the year in which it is offered; acceptance of the Scholarship offer cannot be deferred. If a student declines the Scholarship offer, the Scholarship will be offered to the next eligible student.
The Scholarships shall be administered by the Australian Institute for Machine Learning (AIML)of the University.
The University may vary the rules from time to time in a manner consistent with the University鈥檚 legal obligations and policies.
Applications and enquiries
For general information about the program, please contact Hilary Brookes.
Information on the AIML Summer Research Scholarship Program 2024 to come.
If you have a specific question about a particular project, please contact the listed supervisor(s).
Please apply via the online form.
And there's more
There are six (6) more machine learning opportunities available via the which are centrally administered by University Scholarships. These scholarship projects have different criteria, rules and application requirements.
The projects include:
Automatic poster generation from scientific research papers: Leveraging advanced machine learning techniques
AI-based object detection
Apply a visual foundational model for tumour segmentation
ChatGPT on stock market prediction
Deep inversion for medical image reconstruction using generative adversarial networks
Intelligence-based analysis of sarcoma MRI for predicting histological grade.
