Zahra Dehghanipour

Biography

I am a PhD Candidate in Computer Science at Northeastern University London, researching the scalable verification of deep learning models. My work focuses on building reinforcement-learning-driven branching heuristics and graph-based representations to improve the performance and scalability of branch-and-bound neural network verification.

My research intersects formal verification, trustworthy and explainable AI, and robust machine learning, with an emphasis on improving the reliability and transparency of deep learning systems used in safety-critical environments. I have presented elements of this research at several academic events, including NLDL (Norwegian Machine Learning Conference), Hopper Colloquium (UCL), and Advance HE.

I am also a PhD Student Ambassador, representing the doctoral community and engaging in public-facing communication to highlight the PhD experience at Northeastern University London. 

Qualifications

PhD in Computer Science, Northeastern University London

Thesis: Scalable Verification of Deep Learning Models

Supervisors: Dr Alexandros Koliousis, Dr Elena Botoeva, Prof Alex Freitas 

M.Sc. in Computer Engineering, University of Kashan

Thesis: Dialogue System Development Using Deep Reinforcement Learning

B.Sc. in Computer Engineering, Sheikh Bahaei University

Thesis: Mobile Application for Emergency Medical Requests

Research