I am a Postdoctoral Research Fellow at Applied Artificial Intelligence Institute at Deakin University, Victoria, Australia. I completed my PhD degree from the Centre for Pattern Recognition and Data Analytics group (PRaDA) under the supervision of A/Prof. Santu Rana, A/Prof. Sunil Gupta, Dr. Thin Nguyen and Alfred Deakin Prof. Svetha Venkatesh.

Research Interest

Artificial Intelligence systems are vulnerable to various types of attacks such as Trojan or backdoor attacks, and adversarial attacks. My research interest is to develop mechanisms to understand Adversarial Machine Learning such as adversarial attacks, and backdoor attacks. I focus on building frameworks to defend against such attacks in machine learning models to make them trustable and safe to use. I am also interested in machine unlearning, differential privacy, data manifold analysis, and out of distribution detection.

News

• July 2024: Delivered a talk about AI Security at IEEE Student Branch, Deakin Univeristy.
• July 2024: Our paper “Revisiting the Dataset Bias Problem from a Statistical Perspective” got accepted in ECAI 2024.
• April 2024: Our paper “Composite Concept Extraction through Backdooring” got accepted into Fine-Grained Visual Categorization, CVPR 2024 workshop.
• March 2024: Delivered a keynote talk at the Second International Conference on FOSS Approaches towards Computational Intelligence and Language Technology (FOSS-CIL T24).
• Jan 2024 : Served as a panellist in the Black in AI Emerging Leaders Grad prep program entitled, “MSc/Ph.D: What to expect”.
• Oct 2023 : Invited to be a mentor in Women in Machine Learning (WiML) Workshop 2023.
• July 2023 : Our Workshop, “Backdoors in Deep Learning: The Good, the Bad, and the Ugly” has been accepted to NeurIPS 2023.
• June 2023 : Delivered a talk at MAIL Research Lab at VinUniversity about “Exploring Backdoor Intrusions, Defense Strategies, and the Path to Social Good”.
• September 2022 : Our paper “Momentum Adversarial Distillation: Handling Large Distribution Shifts in Data-Free Knowledge Distillation” has been accepted in NeurIPS 2022.
• July 2022 : Our paper “Towards Effective and Robust Neural Trojan Defenses via Input Filtering” is accepted at ECCV 2022.
• June 2021 : Our paper “Prescriptive analytics with differential privacy” got accepted in International Journal of Data Science and Analytics.
• May 2021 : Our paper “Personalized single-cell networks: a framework to predict the response of any gene to any drug for any patient” in BioData Mining journal.
• June 2020 : Our first paper in Trojan defense “Scalable Backdoor Detection in Neural Networks” got accepted at ECML 2020.