Elena Mocanu

Elena Mocanu is an Assistant Professor in the Department of Datamanagement & Biometrics and a faculty member of the Digital Society Institute. Her research focuses on advancing neural network architectures, particularly through dynamic sparse training techniques to enhance computational efficiency and model performance. Key areas include deep reinforcement learning applications in building energy optimization, federated learning for collaborative data environments, and sparse connectivity models that reduce resource usage without sacrificing accuracy. She has contributed to frameworks like the Digital Twin for autonomous driving and energy systems, emphasizing sustainability and scalability.

Her work bridges theoretical advancements and practical implementations, addressing challenges in energy-efficient AI, robust noise filtering in reinforcement learning, and feature selection for medical imaging tasks. She actively organizes conferences such as ICLR workshops on sparsity in neural networks and IJCAI events, fostering interdisciplinary collaboration in artificial intelligence.

• Awards: Best Paper Award at AAMAS 2022 Workshop, ICML 2022 Outstanding Reviewer Award
• Conference Leadership: Organized ICLR 2023 Sparsity Workshop, EPIA 2023/2022 Conferences

Research highlights include scalable training methods inspired by network science, energy optimization in smart buildings, and sparse ensembling techniques that achieve efficiency gains without overhead. Her contributions span foundational machine learning theory to applied domains like smart grids and autonomous systems.