Scholar
Harleen Hanspal
Google Scholar ID: XMbPUCgAAAAJ
Department of Computing, Imperial College London
Citations & Impact
Citations
4
H-index
1
i10-index
0
Publications
6
Co-authors
9
list available
Publications
Resume (English only)
Academic Achievements
- She has published 'Efficient verification of NNs for LVMs-based Specifications' at CVPR 2023, which developed an approach using Latent Variable Models to improve the efficiency and effectiveness of verifying neural networks' robustness against diverse, data-dependent semantic perturbation sets. Another publication, 'Robustness to Perturbations in the Frequency Domain: Neural Network Verification and Certified Training,' was presented at WACV-W 2023, proposing a framework to encode and verify perturbation sets characterized by their frequency characteristics and suggesting deterministic certified defense against frequency-based attacks.
Research Experience
- Prior to joining Imperial, she worked at Daedalean AG, where she contributed towards developing certifiable autonomous autopilot for manned aerial vehicles.
Education
- From 2017 to 2020, she pursued a Master's degree in Robotics, Systems, and Controls at ETH Zurich, with theses focusing on provable robustness and analysis of Neural Networks, particularly on scaling applications of inexact verification methods, and distributed stochastic optimization under data-driven settings. She obtained her Bachelor's degree in Electrical Engineering from the Indian Institute of Technology Bhubaneswar between 2013 and 2017, with a thesis on Manipulators on Micro Aerial Vehicles.
Background
- Harleen Hanspal is a PhD candidate in the Verification of Autonomous Systems group led by Prof. Alessio Lomuscio at the Department of Computing, Imperial College London. Her research focuses on verifying neural networks against semantic changes in the image domain using generative modeling and defining and verifying semantic properties in language domains, multiagent social settings, etc.
Miscellany
- She enjoys creating consolidated diagrams on topics she studies and shares resources related to ML architectures, operating systems, terminals, UNIX, Docker, PyTorch, and more.
