Ali Harakeh

Papers

Total Citations

H-Index

About

Ali Harakeh is a researcher specializing in deep learning for robotic perception, with a particular focus on uncertainty estimation in neural network-based systems. His most notable contribution is **BayesOD**, a Bayesian framework designed to address one of the critical challenges in deploying deep neural networks within robotic systems: the absence of reliable uncertainty measures associated with model predictions. By integrating Bayesian inference into deep object detection pipelines, Harakeh's work provides a principled approach to quantifying uncertainty, a capability essential for safe and robust autonomous systems operating in unpredictable real-world environments. His research sits at the intersection of probabilistic deep learning and computer vision, tackling the practical limitations of existing uncertainty estimation methods that had previously seen limited success when applied to object detectors. BayesOD has garnered meaningful attention from the robotics and autonomous driving communities, accumulating citations across both its preprint and published versions, reflecting strong interest from researchers grappling with the same reliability challenges. Harakeh's contributions are particularly relevant to the growing field of autonomous vehicles and safety-critical AI, where understanding *when* a model might be wrong is just as important as the predictions themselves.