Find Academic Supervisors

Andrea MontanariView profile

Professor

• High-dimensional statistics
• Machine learning theory
• Statistical physics
• +4 more

Andrea Montanari is a Professor of Mathematics and Statistics at Stanford University, affiliated with the Department of Mathematics and Statistics. His research focuses on high-dimensional statistics, machine learning theory, optimization algorithms, and statistical physics, with applications to neural networks and complex systems. He has contributed extensively to understanding generalization in overparametrized models, spin glass theory, and algorithmic methods like approximate message passing. His work bridges theoretical computer science and mathematical physics, addressing challenges in data analysis and learning from high-dimensional datasets. Notable themes include analyzing neural network dynamics, optimizing high-dimensional landscapes, and developing efficient algorithms for sparse and low-rank matrix estimation. Montanari’s publications explore topics such as the interplay between statistical and computational limits, the behavior of gradient-based methods, and the theoretical foundations of modern machine learning. His recent research demonstrates a focus on fundamental questions in learning theory, including the study of phase transitions in statistical estimation, the role of overparametrization in generalization, and the mathematical underpinnings of contemporary algorithms. While no specific awards are listed here, his contributions reflect significant impact in interdisciplinary fields.

Anna Dawid-LekowskaView profile

Assistant Professor

• Interpretable Machine Learning
• Explainable Artificial Intelligence
• Quantum Simulation
• +7 more

Anna Dawid-Lekowska is an Assistant Professor at the Leiden Institute of Advanced Computer Science (LIACS) and affiliated with the Leiden Institute of Physics (LION) at Leiden University, Netherlands. She leads a research group within the aQa group, focusing on the intersection of machine learning and quantum physics. Previously, she was a Research Fellow at the Center for Computational Quantum Physics, Flatiron Institute, New York. PhD in Physics and Photonics (joint, cotutelle), University of Warsaw & ICFO, Spain MSc in Quantum Chemistry, University of Warsaw BSc in Biotechnology, University of Warsaw Anna's research centers on interpretable machine learning for scientific discovery, particularly in quantum systems. She investigates how overparametrized models generalize, the role of loss landscape flatness, and double descent phenomena. Her work bridges deep learning with quantum simulations, aiming to detect quantum phase transitions and extract physical insights from trained models. She also explores ultracold molecules and novel quantum phases using simulation platforms. Her recent publications demonstrate a strong trend in applying machine learning to automate and interpret quantum experiments, such as detecting laser cooling schemes and understanding neural network initialization. The work emphasizes interpretability, aiming to make AI a transparent scientific tool rather than a black box. Anna has received significant recognition, including: 2022 FNP START laureate Participant in the 2024 Lindau Nobel Laureate Meeting She is actively mentoring and expanding her group, currently recruiting PhD students and postdoctoral researchers. Her work is supported by institutional affiliations with leading research centers and collaborations across Europe and the US. Anna also engages in science communication and education, having lectured at the Nordita Winter School on Machine Learning and Physics. She is involved in the aQa research group, which focuses on quantum algorithms and AI, fostering interdisciplinary collaboration between computer science and physics. Her lab integrates theoretical modeling, algorithm development, and applications to quantum experiments.

Eren C KızıldağView profile

Assistant Professor

• Statistics
• Data Science
• Probability Theory
• +5 more

Eren C. Kızıldağ is an Assistant Professor in the Department of Statistics at the University of Illinois Urbana-Champaign, with additional affiliations in the Department of Electrical and Computer Engineering. He holds a PhD in Electrical Engineering and Computer Science from MIT, where he was part of the Laboratory for Information and Decision Systems (LIDS) and the Institute for Data, Systems, and Society (IDSS). Previously, he was a Distinguished Postdoctoral Fellow at Columbia University. His research bridges probability, statistics, and computer science, focusing on statistical-computational trade-offs in random models such as optimization problems and statistical inference. Key interests include understanding algorithmic barriers in high-dimensional problems, discrepancy theory, and neural network theory. His work often explores connections to statistical physics and combinatorial structures. Recent publications highlight contributions to topics like low-rank tensor recovery, algorithmic obstructions in partitioning problems, and geometric barriers in discrepancy minimization. His research also addresses theoretical foundations of overparameterized neural networks and hardness results for partition functions in spin glass models. Kızıldağ’s academic journey includes a Master’s from MIT and a B.S. (summa cum laude) from Bogaziçi University, with early work in MRI technology at ISMRM. His research is supported by collaborations with institutions like LIDS and IDSS, and his work frequently appears in top venues such as Annals of Applied Probability, Mathematics of Operations Research, and IEEE ISIT.

Kangjie ZhouView profile

Research Fellow

• High-dimensional statistics
• Non-convex optimization
• Probability theory
• +3 more

Kangjie Zhou is a Founder's Postdoctoral Fellow in the Department of Statistics at Columbia University. He completed his PhD in Statistics at Stanford University under the advisement of Professor Andrea Montanari, with additional collaborative work with Professor Tselil Schramm. He holds a Bachelor's degree in Mathematics from Peking University's School of Mathematical Sciences. His research focuses on high-dimensional statistics, non-convex optimization, probability theory, and deep learning. Notable contributions include work on computational-to-statistical gaps, theoretical aspects of deep learning, and high-dimensional probability. He has been recognized with the 2024 Department Dissertation Award. Key research themes include analysis of tensor decompositions, optimization landscapes in neural networks, and the interplay between overparametrization and computational tractability. His work bridges statistical theory with algorithmic development, often addressing foundational questions in machine learning and data science.

Tianyang HuView profile

Assistant Professor

• Statistical Machine Learning
• Representation Learning
• Deep Generative Modeling
• +7 more

Tianyang Hu is a postdoctoral researcher at the National University of Singapore, set to join the Chinese University of Hong Kong, Shenzhen as an Assistant Professor in Fall 2025. His academic journey includes a PhD in Statistics from Purdue University under Prof. Guang Cheng, an M.S. in Statistics from the University of Chicago, and a B.S. in Mathematics from Tsinghua University. PhD in Statistics, Purdue University M.S. in Statistics, University of Chicago B.S. in Mathematics, Tsinghua University His research lies at the intersection of statistics and artificial intelligence, focusing on statistical machine learning, representation learning, and deep generative modeling . He is particularly interested in the theoretical foundations of AI, including diffusion models, large language models (LLMs), and overparametrized neural networks. His recent work explores the duality between prompting and fine-tuning, contrastive learning mechanisms, and efficient sampling in generative models. The trends in his recent publications show a strong emphasis on diffusion models , especially in conditional generation, sampling efficiency, and integration with language models. His work spans both theoretical analysis and practical algorithm design, with applications in image generation, 3D modeling, and multimodal reasoning. He has contributed to understanding the behavior of in-context learning in transformers and the role of regularization in deep networks. His scientific achievements include: Spotlight presentation at NeurIPS 2023 Spotlight presentation at NeurIPS 2022 Oral presentation at ICML Workshop 2025 Tianyang Hu is actively building his research group and is seeking PhD/MPhil students and research assistants with strong theoretical and practical AI skills. He has held research positions at Huawei Noah’s Ark Lab and is currently mentored by Prof. Kenji Kawaguchi at NUS. His work has been supported by collaborations across academia and industry, though specific grants are not listed in the text. While no formal lab name is mentioned, his future group at CUHK-Shenzhen is expected to focus on theoretical AI and generative models , with research directions in LLMs, diffusion models, and representation learning.

René VidalView profile

Professor

• Machine Learning
• Computer Vision
• Biomedical Data Science
• +3 more

René Vidal is the Rachleff & Penn Integrates Knowledge (PIK) University Professor at the University of Pennsylvania, with appointments in the Departments of Electrical and Systems Engineering, Radiology, Computer and Information Science, and Statistics and Data Science. He also serves as Director of the Center for Innovation in Data Engineering and Science (IDEAS) and the NSF-Simons Collaboration on the Mathematical Foundations of Deep Learning (THEORINET). A dual faculty member at Johns Hopkins University in Biomedical Engineering, Computer Science, and other departments, Vidal is an Amazon Scholar and Affiliated Chief Scientist at NORCE. PhD, Electrical Engineering and Computer Sciences, University of California, Berkeley (2003) M.S., Electrical Engineering, University of California, Berkeley (2000) B.S. (valedictorian), Electrical Engineering, Pontificia Universidad Catolica de Chile (1997) His research spans the mathematical foundations of deep learning, focusing on non-convex optimization , learning dynamics , and overparametrization . Key contributions include Sparse Subspace Clustering , Kernel GPCA , and Low-Rank Matrix Factorization , with applications in autism diagnosis , robotic surgery , and cardiac imaging . Recent work explores continual learning , adversarial robustness , and trustworthy AI in biomedical contexts. His 15 most recent publications highlight advances in medical imaging , language models , and robust computer vision , spanning topics from chest X-ray analysis to motor imitation tasks in autism . Articles like Geometric Analysis of Nonlinear Manifold Clustering underscore his theoretical contributions, while works on KDA: Knowledge-Distilled Attacker and Conformal Information Pursuit address practical AI safety and interpretability. Scientific accolades include: 2022 ACM Fellow 2021 IEEE McCluskey Technical Achievement Award 2017 Jean D’Alembert Fellowship 2012 J.K. Aggarwal Prize 2009 Sloan Research Fellow 2005 NSF CAREER Award His lab mentors 11 current PhD students across Johns Hopkins and University of Pennsylvania , with alumni contributing to institutions like Meta , Amazon , and GE Research . Vidal’s interdisciplinary work bridges mathematics , engineering , and healthcare , supported by grants from the DoD , NSF , and ONR .

• Machine Learning Theory
• Statistical Physics
• Neuroscience
• +5 more

Stefano Sarao Mannelli is a tenure-track Assistant Professor in the Department of Computer Science and Engineering at Chalmers University of Technology and University of Gothenburg. He also holds a Visiting Lecturer position at the University of the Witwatersrand. His research group focuses on fundamental aspects of learning in biological and artificial systems, with emphasis on bias generation, optimization dynamics, and comparative neuroscience. Education: Ph.D. in Theoretical Physics, Université Paris-Saclay (2020) M.Sc. in Electronic Engineering, Politecnico di Torino (2017) M.Sc. in Physics of Complex Systems, Politecnico di Torino/SISSA (2016) M2 in Physique Théorique, Paris Diderot/UPMC/ENS Cachan (2016) B.Sc. in Mathematics for Engineering, Politecnico di Torino (2014) Research: Dr. Mannelli develops model-based approaches to reduce complex machine learning problems into analytically tractable frameworks. His core interests include: 1) Bias amplification mechanisms in AI systems, 2) Learning differences between biological and artificial neural networks (continual/transfer/curriculum learning), and 3) Optimization in high-dimensional landscapes. His work bridges statistical physics, neuroscience, and deep learning theory. Publication Trends: Recent articles (2024-2025) predominantly analyze curriculum learning dynamics, bias propagation in optimization, and theoretical comparisons between biological and artificial learning systems. Methodologically, they combine statistical physics frameworks with control theory and high-dimensional analysis. Awards: Academic Grant (CM Lerici Foundation, 2025) Travel Grants (Guarantor of Brains, G-Research 2024) UK–IT Trustworthy AI Exchange Programme (Alan Turing Institute, 2023) SCGB Conference Award (Simons Foundation, 2023) Ph.D. Scholarship (CEA, 2017-2020) Team & Funding: Leads a research group with 2 PhD students and 1 postdoc. Secured significant funding for international workshops including Analytical Connectionism (£42K, 2023; $152K, 2024) and High-Dimensional Methods (135,500 SEK, 2025).

• Mathematical Statistics
• Machine Learning
• Kernel Methods
• +3 more

Prof. Dr. Nicole Mücke is a Professor in the Institute for Mathematical Stochastics at the Carl-Friedrich-Gauss Faculty of Technische Universität Braunschweig. Her research focuses on mathematical statistics, machine learning, kernel methods, and statistical inference. She explores topics such as neural network theory, inverse problems, optimization, and regularization techniques. Her work bridges theoretical foundations with practical applications in areas like distributed computing and uncertainty quantification. Prof. Mücke’s research portfolio includes contributions to empirical risk minimization, neural operator learning, and gradient-based optimization. She investigates the interplay between overparameterization and generalization in machine learning models, as well as the design of efficient algorithms for large-scale problems. Her publications span topics ranging from distributed stochastic gradient descent to localized kernel regression techniques. Her recent work emphasizes theoretical guarantees for learning algorithms, including convergence rates, statistical performance in high-dimensional settings, and the role of regularization in inverse problems. She also explores methodological advancements in spectral methods, algorithm unfolding, and data-splitting strategies to enhance statistical efficiency. Prof. Mücke’s research is characterized by a strong emphasis on rigorously analyzing machine learning algorithms through the lens of statistical theory and functional analysis. Her contributions address challenges in both classical and modern machine learning paradigms, with a focus on bridging the gap between abstract mathematical frameworks and practical implementation.

Damek DavisView profile

Visiting Associate Professor

• Optimization
• Machine Learning
• Signal Processing
• +4 more

Damek Davis is a Visiting Associate Professor at the Wharton Department of Statistics and Data Science at the University of Pennsylvania and an Associate Professor of Operations Research at Cornell University (currently on leave). He held NSF Postdoctoral Fellowships and completed his PhD in Mathematics at UCLA under Wotao Yin and Stefano Soatto. His research bridges optimization , machine learning , statistics , and signal processing , with a focus on convergence guarantees for stochastic and nonsmooth optimization methods. He has developed accelerated gradient descent algorithms, characterized SGD behavior in nonconvex settings, and formalized avoidance of strict saddle points in proximal methods. Key publications include exponential accelerations of gradient-based techniques and first guarantees for SGD on weakly convex functions. His work has been recognized by the Sloan Research Fellowship INFORMS Optimization Society Young Researchers Prize NSF CAREER Award SIAM Activity Group on Optimization Best Paper Prize Davis actively advises Penn graduate students and serves as an associate editor at Mathematical Programming and Foundations of Computational Mathematics . He emphasizes clear technical writing and public communication, maintaining a blog and lecture notes on optimization theory.

Alexander RakhlinView profile

Professor

• Machine Learning
• Statistics
• Optimization
• +3 more

Alexander Rakhlin is a Professor in the Department of Brain & Cognitive Sciences at MIT, with affiliations at the MIT Statistics and Data Science Center and the Institute for Data, Systems, and Society (IDSS). His research bridges statistics, machine learning, and optimization, focusing on online prediction, reinforcement learning, and decision-making under uncertainty. He holds a PhD from MIT (2006) and previously served as an associate professor at the University of Pennsylvania (Wharton School) before joining MIT in 2023. Research interests include reinforcement learning, neural networks, overparametrized models, and the theoretical foundations of machine learning. His work addresses challenges in sequential decision-making, robust algorithms, and the statistical complexity of modern models like large language systems. Notable contributions include the Decision-Estimation Coefficient framework for RL analysis and insights into benign overfitting in overparametrized settings. Recent publications emphasize algorithmic advances in online learning, privacy-aware contextual bandits, and structural watermarking for language models. His theoretical work often unifies principles from information theory, optimization, and statistical learning. Lab affiliations include the Laboratory for Information and Decision Systems (LIDS), where he collaborates on interdisciplinary projects. No specific awards or grants are explicitly listed in the provided texts.

• Machine Learning
• Optimization
• Deep Learning Theory
• +7 more

Prof. Dr. Aurelien Lucchi is an Assistant Professor in the Department of Mathematics and Computer Science at the University of Basel, Faculty of Science. His research group focuses on the intersection of optimization and machine learning, particularly in advancing theoretical understanding and algorithmic design for deep learning systems. His research interests include: Stochastic and non-convex optimization Deep learning theory and generalization Kernel methods and spectral analysis Transformer architectures and training dynamics Batch Normalization and initialization effects Modeling optimization via stochastic differential equations (SDEs) The recent publications (2023–2025) highlight a strong focus on theoretical machine learning, especially in characterizing optimization landscapes, generalization in kernel methods, and the role of noise and adaptive methods in training. There is a clear trend toward using advanced mathematical tools—such as random matrix theory, SDEs, and curvature analysis—to explain phenomena in deep learning. His group actively publishes in top venues including NeurIPS, ICML, ICLR, and AISTATS. Scientific awards and recognitions include: SNF Consolidator Grant (1.7M CHF) Prof. Lucchi leads an active research group with postdoctoral fellows and ongoing projects, including work on quantum machine learning and noise-adaptive optimization. He has secured competitive research funding and mentors early-career researchers. His group has received recent paper acceptances at ICLR 2025, AISTATS 2025 (oral), and NeurIPS 2024, indicating strong momentum in theoretical and algorithmic machine learning. He previously held a scientific research position at ETH Zurich (2014–2021) and earned his PhD from EPFL. The group is currently involved in two major ongoing projects: Designing and Training Hybrid Hierarchical Quantum Neural Networks with Quantum Advantage Noise-Adaptive Optimization Methods and their Robustness Properties

• Mathematical Foundation of Deep Learning
• Deep Learning Theory
• Optimization
• +3 more

Chaoyue Liu is an Assistant Professor in the Elmore Family School of Electrical and Computer Engineering at Purdue University, located in West Lafayette. His research focuses on the mathematical foundations of deep learning, deep learning theory, and optimization techniques. He holds a position in the Department of Electrical and Computer Engineering and contributes to advancing theoretical and practical aspects of machine learning systems. Research Interests: Mathematical underpinnings of deep learning architectures and their theoretical guarantees Optimization algorithms for large-scale machine learning models Federated learning and privacy-preserving techniques Analysis of neural network training dynamics and loss landscapes Key Research Trends in Publications: Recent work emphasizes optimization challenges in over-parameterized systems, federated learning frameworks, and the dynamics of neural network training. His articles explore topics like SGD batch saturation, differential privacy integration, and the emergence of linear behavior in wide neural networks. Grants & Advising: No specific grants or student advisement information is listed. Collaborations may involve Purdue's research groups in machine learning and electrical engineering. Labs/Teams: Affiliated with Purdue's Electrical and Computer Engineering research clusters, likely contributing to interdisciplinary projects in AI and signal processing.

Omri WeinsteinView profile

Assistant Professor

• Theoretical Computer Science
• Data Structures
• Information Theory
• +12 more

Omri Weinstein is an Assistant Professor in the Department of Computer Science at Columbia University. His research bridges Information Theory, Data Structures, and Optimization, focusing on dynamic data structures and dimensionality-reduction techniques to accelerate optimization and search. He received his PhD from Princeton University and was a Simons Society Junior Fellow at the Courant Institute (NYU). Education: PhD in Computer Science, Princeton University Simons Society Junior Fellowship, Courant Institute (NYU) His work explores the theoretical foundations of data structure lower bounds, communication complexity, and secure computation. Recent research includes advancements in dynamic matrix inversion for linear programming, oblivious near-neighbor search, and the interplay between matrix rigidity and data structure efficiency. Key trends in his publications include: Proving polynomial and super-logarithmic lower bounds for static and dynamic data structures Developing novel techniques in information complexity and protocol compression Applications in parallel algorithms, compressed data structures, and algorithmic game theory Scientific Awards: NSF CAREER Award Simons Society Junior Fellow Best Paper Award at CSR '13 Advising and Grants: Advised PhD students Hengjie Zhang and Shunhua Jiang MsC student Victor Lecomte and postdoc Alexander Golovnev Research funded by NSF CAREER Award on data structure lower bounds Labs and Teams: Omri is affiliated with the Theoretical Computer Science Group at Columbia and leads the Data-Structure Lower Bounds Reading Group.

Antonio OrvietoView profile

Lecturer

• Deep Learning
• Optimization Theory
• Neural Network Architecture Design
• +7 more

Antonio Orvieto is a Lecturer at the University of Tübingen, Principal Investigator at the ELLIS Institute Tübingen, and independent group leader at the Max Planck Institute for Intelligent Systems. He leads the Deep Models and Optimization research group and serves as faculty for the CLS, ELLIS, and IMPRS-IS PhD Programs. Orvieto holds a PhD from ETH Zürich and has conducted research at DeepMind London, Meta (FAIR) Seattle, MILA, and Inria Paris. Orvieto's research focuses on improving the efficiency and accessibility of deep learning technologies through theoretical advancements in optimization and architecture design. His work spans two main areas: understanding large-scale optimization dynamics and designing innovative neural network architectures capable of reasoning with complex sequential data. His research has significant implications across multiple domains including biology, neuroscience, natural language processing, and music generation. His approach combines rigorous theoretical analysis with practical applications to address fundamental challenges in deep learning. His recent publications reveal a strong emphasis on understanding optimization landscapes, developing efficient recurrent architectures, analyzing transformer behavior, and exploring the theoretical foundations of state-space models. The work shows a consistent pattern of bridging theoretical insights with practical implementations, particularly in handling sequential data and improving training efficiency. Schmidt Sciences AI2050 Early Career Fellow Orvieto actively mentors PhD students and researchers in his Deep Models and Optimization group, which includes PhD candidates working on various aspects of deep learning theory and applications. His research is supported through his positions at the University of Tübingen, ELLIS Institute, and Max Planck Institute for Intelligent Systems. He has collaborated with leading researchers across multiple institutions including ETH Zürich, DeepMind, Meta, MILA, and Inria. His research group focuses on investigating the interplay between optimizers and architectures in deep learning, with particular emphasis on developing new networks for long-range reasoning. The group strongly believes that deep learning will revolutionize science and technology, and they aim to provide theoretical foundations that will enable scientists and engineers with limited resources to leverage powerful deep learning solutions.

Enrique MalladaView profile

Associate Professor

• Control of Networked Systems
• Power Systems
• Optimization
• +5 more

Enrique Mallada is an Associate Professor of Electrical and Computer Engineering at Johns Hopkins University (JHU), with secondary appointments in Mechanical Engineering, Applied Mathematics and Statistics, and Computer Science. He leads the Networks, Dynamics, and Learning Laboratory (NetDL2ab) and is a core member of MINDS (Mathematical Institute for Data Science) and ROSEI (Ralph O’Conner Sustainable Energy Institute). His research focuses on control systems, optimization, power systems, and machine learning, with applications in networked systems, energy grids, and distributed coordination. Education: B.S. in Telecommunications Engineering from Universidad ORT (2005), Ph.D. in Electrical and Computer Engineering with a minor in Applied Mathematics from Cornell University (2014). Postdoctoral research at Caltech’s Center for the Mathematics of Information (2013–2015). Research Interests : Networked systems (synchronization, distributed coordination), power systems (frequency control, grid resilience), optimization (time-varying algorithms, reinforcement learning), and machine learning (safety-critical applications, sparse recovery). Key Projects : Real-time optimization for infrastructure networks, voltage collapse stabilization in power grids, control of distributed energy resources, and safety-aware reinforcement learning. His work integrates tools from control theory, optimization, and machine learning to address challenges in large-scale systems. Awards : NSF CAREER Award, Caltech CMI Fellowship, Cornell Jacobs Fellowship, JHU Discovery/Catalyst Awards, and Excellence in Teaching Award. Advising & Grants : Mentors graduate students in ECE and related fields. Active in grants focused on energy systems, control theory, and AI safety. Organizes conferences like CISS 2019 and serves on technical committees for IEEE Smart-GridComm and ACC. Labs & Affiliations : NetDL2ab Lab (directed), MINDS (core member), ROSEI (core member), LCSR (affiliate), and Data Science & AI Institute (member). His research bridges academia and industry, addressing real-world challenges in energy and automation.