Applying generative neural networks to automatically create realistic looking faces from skulls with no human interaction and in real-time.

Determining the location of electrical failures in the power grid

Using AI techniques to help researchers track the health of our ecosystem by monitoring recorded bioacoustics data in the wild.

Handling crowd navigation for robots without getting lost or colliding with anyone, but also behave in a socially confirmative way such that the robot moves in a predictable way to avoid making humans stop, slow down, or alter their intended path of travel.

Generative deep learning models look at learning data distributions to be able to reconstruct similar types of data from scratch.

In collaboration with Dr Erika Gyengesi and the Gyengesi Lab from Western Sydney University (WSU), the Microglia Analysis project aims to develop a deep learning framework that uses microglial morphology to characterise and standardise disease states, potentially aiding research into neurodegenerative conditions such as dementia and alzheimer’s disease.

The Microfluidics project is dedicated to optimising the identification of viable sperm concentrations in IVF treatment. Presently, a labour-intensive process requires a skilled clinician with extensive training to manually select and evaluate samples. Through the implementation of advanced Deep Learning techniques, the project seeks to streamline this procedure, potentially reducing the time commitment from skilled clinicians to mere seconds. This transformative approach not only enhances efficiency but may also represent a significant leap forward in reproductive healthcare.

In partnership with Yale and the Global Meded Network, the Ultrasound Project endeavours to integrate AI models as diagnostic tools for General Practitioners, specifically focusing on cardiac ultrasound diagnosis in resource-constrained areas. This initiative has been successfully deployed in Haiti, where it is actively utilised by local General Practitioners. By leveraging cutting-edge technology and collaborative expertise, the Ultrasound Project addresses healthcare challenges in underserved regions, enhancing diagnostic capabilities and contributing to the advancement of medical care in low-resource settings.

In partnership with Monash High Powered Rocketry (MHPR), the project is dedicated to the development of a Reinforcement Learning (RL) framework for precise rocket control in the lander challenge. The objective is to achieve a controlled ascent, hover, and precise landing. Through the collaboration with MHPR, we aim to pioneer advancements in rocketry technology, employing sophisticated RL techniques to meet the exacting requirements of this challenge and contribute to the evolution of precision landing capabilities in aerospace engineering.

Representing Monash DeepNeuron on the world stage, the DL competitions team features a collection of Deep Learning (DL) members with varying degrees of experience who work together to develop solutions for select AI competitions throughout the semester. These competitions investigate exciting areas of AI research through challenging use-cases, which present an opportunity for members to learn and explore new DL techniques and solidify their DL skillset. In S2 2023, the team competed in the Google American Sign Language Fingerspelling Recognition and CommonLit Evaluate Student Summaries competitions.

This project involves the digitisation of real-world objects into viewable/interactive three dimensional models, allowing users to scroll around and view all angles in a digital format as if they had the object in front of them. This project is a collaboration with Monash Automation, another student engineering team, who are working on the creation of robots that can take videos and images of objects which will be digitised.

Exploring an exciting field of AI research, the Image Generation using Graph Neural Networks (GNNs) project applies graph theory and explores the usage of deep GNN models for extracting semantic information of objects and their relationships from graph representations of images. Through the application of these techniques, this project aims to develop an end-to-end image generation pipeline that provides complete control over object semantics, contributing to the niche area of research that is GNNs.