I am David D. Nguyen, a researcher in artificial intelligence and machine learning, with a particular focus on generative models and their applications in cybersecurity. My PhD was completed at the University of New South Wales under the guidance of Professor Salil Kanhere (UNSW), Dr. Surya Nepal (CSIRO) and Dr. David Liebowitz (Penten). I have published peer-reviewed papers on various generative modeling problems, including images, layouts, software code, and documents, in top computer science venues such as AAAI, ACM Multimedia, and HICSS.

Currently I am at CSIRO’s Data61, Australia’s National Research Laboratory, specializing in machine learning problems within the cyber security domain. My recent research interests are denoising diffusion models, density estimation, adversarial purification, adversarial robustness and text-to-speech recognition. If you are interested in collaborating or a PhD research position at CSIRO, please feel free to contact me at (d.nguyen AT csiro.au).

Recent News

10/12/23 - Paper accepted into AAAI 2024.

Our latest paper introduces Multiple Hypothesis Dropout (MH-Dropout), a novel variant of dropout that converts a single-output function into a multi-output function. Check out our paper here and code here! The illustration below depicts a Mixture of Multiple Output functions (MoM).

This novel technique employ subnetworks from a base neural network to estimate the parameters of multi-modal Gaussian distributions. Using this technique, we propose an improved VQGAN that generates higher quality images using a significantly smaller codebook. The samples below were generating using only 4-64 codebook entries.

22/5/22 - Best paper award from UNSW.

Very honoured to have won the Norman Foo Memorial Best PhD Paper Award from UNSW! Check out the paper here which introduces a model that generate layouts with a new multiple choice learning technique that avoids modal collapse and prediction averaging.