Adel Elzemity

Enhancing the Security and Privacy of AI Models in a Dynamic Threat Landscape

This research investigates the security and privacy risks of AI models, focusing on large language models (LLMs) and federated learning (FL). It explores how fine-tuning affects AI safety, the vulnerabilities in decentralised learning, and methods to improve security while maintaining performance.  

AI Security & Privacy
Adversarial Attacks & Defences in AI
Model Fine-Tuning & Security Trade-Offs  

As a GTA, I have been (still am) assisting in teaching the following modules:

Previously:
COMP8760
COMP8240
COMP6644/8340
COMP5280
COMP6570

This Term:
COMP6685
COMP5580
COMP6685  

Budi Arief and Shujun Li