Research
Research
We are generally interested in topics where neural networks are deployed in resource-constrained environments (e.g. edge devices, robotic systems), and AI tools interface with human users for sensitive decision-making (e.g. problem solving, policy decisions). We also apply machine learning to identify complex patterns that are predictive of outcomes in business and medicine.
Knowledge Distillation and Representational Alignment
We explore transferring capabilities from larger to smaller models by explicitly utilizing representational alignment measures (P. Bhattarai, et al., 2024). We also explore model architectures that learn closed-form representations across a variety of modalities (images, physiologic signals) in the form of “Feature Imitating Networks” (S. Sadiya et al., 2022; S. Min, et al., 2024; C. Wu, et al., 2025) and develop learnable layers within a neural network to enhance the predictive power of learned representations (N. Eghbali, et al., 2025). The development of such methods provide more efficient training of neural network models which are critical for resource-constrained contexts (e.g. edge devices, energy-sensitive eco-systems).
Machine Learning for AI-driven Neuro-robotics
We study the roles of functional muscle networks (C. Armanini, et al., 2024) and feature imitating networks (C. Wu, et al., 2025) to improve EMG-based hand gesture recognition. These works have significant implications for improving myoelectric control in neurorobots and interactive robots, making such systems more precise and responsive.
Human-Machine Interaction and Decision-making
Wd study human-AI interactions, specifically how AI-generated explanations (factual and counterfactual) influence human decision-making (L. Ibrahim et al., AAMAS 2023), and how user preferences bias decision-making systems within societal and policy-driven contexts (L. Ibrahim et al., HCOMP 2021). We explore how people understand, trust, and utilize AI systems, contributing to the broader theme of improving human decisions through AI assistance.
Augmenting Human Knowledge Acquisition and Problem Solving via LLM Tools
We explore how eye-tracking technology can be leveraged to understand the behavior and decision-making processes of students while programming with the help of a large language model (LLM) assistant. The study provides insights into cognitive load, attention distribution, and problem-solving strategies, highlighting how students interact with AI-driven tools in real-time programming tasks. We also contribute to a broader understanding of the impact of AI tools in education (H. Ibrahim et al., Nature 2023).
The Automated Venture Capitalist
An incredible number of factors influence the success or failure of ventures. Some of these factors are within a venture’s control, and others are not. Our work analyzes factors such as team profiles (Thirupathi et al., 2020), funding (R. Khanmohammadi et al., IEEE TCSS 2024), scientific / patent activities (R. Khanmohammadi et al., 2024), and the complex interactions between them to predict venture success.
Partners
We partner with researchers within the NYUAD Center for AI & Robotics (CAIR) on signal processing, machine learning, and neuro-robotics. We also partner with researchers in the NYUAD Center for Quantum Computing & Topological Systems (CQTS) on quantum machine learning (L. Mecharbat et al., 2025), and machine learning applied to NMR spectroscopy and quantum chemistry.
Publications
For our latest publications, please visit Google Scholar.
Awards
(June 2022) Best Student Paper Award (Nominee), ACM/IEEE JCDL.
Media
(Aug 2023) ChatGPT Can Get Good Grades. What Should Educators Do about It? (Scientific American; New Scientist)
(Oct 2022) Engineering Insight into Mental Health (NYUAD News)