Menu
Explore a research direction
Machine learning theory
research labs.
Explore labs that list machine learning theory among their research interests, including work on the mathematical foundations of learning, optimization and decision making. Each profile links to its official sources so you can check what a group actually studies before you decide whom to contact.
What to compare
When comparing groups in machine learning theory, look at whether the work proves guarantees, designs algorithms or tests them empirically, and what mathematical background is expected. Ask how a new researcher would be involved and what preparation is expected.
Profiles appear here because their listed research keywords match this topic. Inclusion does not establish an open position. Check the official sources and any published application instructions before reaching out.
Prepare a research introduction ↗22 research-led lab profiles
Browse labs by their listed college affiliationAleksander Mądry Research Group
Massachusetts Institute of TechnologyAleksander Mądry · Cadence Design Systems Professor of Computing
How can machine learning become a decision-making toolkit that is reliable and well-understood enough to be deployed safely in the real world? Aleksander Mądry is the Cadence Design Systems Professor of Computing in the MIT EECS Department and a member of CSAIL. His research spans machine learning, optimization and algorithmic graph theory. He is the Director of the MIT Center for Deployable Machine Learning and a Faculty Co-Lead of the MIT AI Policy Forum. His page says he is currently on leave from MIT and spending that time at OpenAI.
Animesh Garg Research Group
Georgia Institute of TechnologyAnimesh Garg · Stephen Fleming Early Career Professor
How can robots gain generalizable autonomy through representations and algorithms for reinforcement learning, control, and perception? Animesh Garg leads the People, AI, and Robotics research group. He is on the core faculty in the Robotics and Machine Learning programs. He currently studies learning structured inductive biases in sequential decision making, using data-driven causal discovery, and transfer to real robots, all in the purview of embodied systems. Animesh is also a Senior Researcher at Nvidia Research.
Computational Biology Laboratory
George Mason UniversityAmarda Shehu · Professor of Computer Science
How can algorithms help explain a molecule’s structure, motion, and function? Amarda Shehu studies computational structural biology, biophysics, and bioinformatics, including probabilistic search and optimization for protein modeling. Her broader AI research develops frameworks for understanding complex biological and engineered systems, with projects involving protein language models and property-controlled molecule generation. The work connects foundational machine learning with scientific questions about biomolecules. It offers a concrete research match for interests in AI for science, molecular modeling, and algorithm design.
Dimitris Bertsimas Research Group
Massachusetts Institute of TechnologyDimitris Bertsimas · Boeing Leaders for Global Operations Professor of Management
How can optimization, stochastic systems and machine learning be applied to practical problems? Dimitris Bertsimas is the Boeing Leaders for Global Operations Professor of Management and a Professor of Operations Research. He has worked in robust optimization, statistics, healthcare, transportation and finance. He has coauthored more than 200 scientific papers. He has supervised 59 doctoral and 31 Master students. He was named Vice Provost for Open Learning in September 2024.
Elad Hazan Research Group
Princeton UniversityElad Hazan · Professor
How can algorithms be designed with provable efficiency and robustness for large-scale learning, online decision-making, and control? The group researches the design and analysis of algorithms for machine learning and optimization, including adaptive gradient methods like AdaGrad and first sublinear-time algorithms for convex optimization. Researchers develop and analyze projection-free and fast semidefinite programming solvers and online Newton-style methods to obtain logarithmic regret guarantees for online convex optimization. Work also introduces nonstochastic control theory linking optimization and control, with implementations and theoretical analysis to enable provable performance.
Gauri Joshi Research Group
Carnegie Mellon UniversityGauri Joshi · Professor
How can scalable and efficient algorithms be designed for distributed machine learning training and inference? Gauri Joshi is a professor in the Electrical and Computer Engineering department at Carnegie Mellon University. She advises a group of students in the Optimization, Probability and Learning lab. Her additional affiliations include the Robotics Institute, the Machine Learning Dept, the Parallel Data Lab and CyLab. One research theme seeks to understand when launching replicas of a task on multiple servers can outweigh the cost of additional resources.
George Konidaris Research Group
Brown UniversityGeorge Konidaris · Professor of Computer Science
Can robots autonomously learn abstract skills that enable fast, goal-directed planning across diverse tasks and environments? The Intelligent Robot Lab constructs high-level symbolic representations and abstraction hierarchies to enable planning, for example work on "From Skills to Symbols" and related projects. The lab develops hierarchical reinforcement learning and probabilistic learning methods to learn task-level skills and then composes them for long-horizon planning using projects such as "Autonomous Robot Skill Acquisition." Researchers integrate optimal control, planning, and learned skills to produce robust behavior under uncertainty in manipulation and navigation. The lab builds applied systems including robot motion-planning hardware and software prototypes described as "Robot Motion Planning on a Chip.
Jenna Wiens Research Group
University of Michigan-Ann ArborJenna Wiens · Professor
How can hospital data support useful clinical decisions when measurements are incomplete and model predictions can be unreliable? The group develops machine-learning approaches using patient images, laboratory tests and vital signs. Research includes predicting infections and other patient outcomes, learning treatment policies with causal inference and offline reinforcement learning, and evaluating how clinicians respond to AI recommendations. Another direction investigates selective prediction: withholding potentially unreliable outputs to reduce automation bias. The work connects algorithm design with the practical effects of deploying models in clinical workflows.
Jonathan Cagan Research Group
Carnegie Mellon UniversityJonathan Cagan · George Tallman and Florence Barrett Ladd University Professor, Mechanical Engineering
How can computational models and AI improve human design problem solving in engineering practice? The Integrated Design Innovation Group researches engineering design automation by merging AI, machine learning, optimization, and cognitive-science models of designer processes. Projects study hybrid human/AI teams, computational methods for biomechanical system design and diagnosis, and user-centered integrated product development. Researchers develop algorithmic agents, multi-agent simulations, and optimization techniques to represent designer cognition and to assist or evaluate real-time design decision processes. The group combines behavioral experiments, computational modeling, and applied engineering datasets to test and refine computational design methods.
Jorge Nocedal Research Group
Northwestern UniversityJorge Nocedal · Walter P. Murphy Professor of Industrial Engineering and Management Sciences
How can optimization algorithms be designed to handle large scale problems with millions of variables? Jorge Nocedal's main area of research is optimization, with applications in machine learning, engineering design, and the physical sciences. His research activities range from the design of new algorithms to their software implementation and mathematical analysis. Areas of emphasis on his page include optimization under uncertainty and parallel computing, alongside large scale problems. He is also the director of the Center for Optimization and Statistical Learning at Northwestern.
Justin K. Romberg Research Group
Georgia Institute of TechnologyJustin K. Romberg · Schlumberger Professor
How can theoretical frameworks and algorithms make data acquisition and inference more efficient? Justin Romberg is the Schlumberger Professor and the Associate Chair for Research in the School of Electrical and Computer Engineering. His research interests lie on the intersection of signal processing, machine learning, optimization, and applied probability. His work investigates structured signal models, compressed sensing, and computational methods for high-dimensional data analysis. He is the Associate Director for the Center for Machine Learning at Georgia Tech.
Justin Romberg Research Group
Georgia Institute of TechnologyJustin Romberg · Schlumberger Professor
How can signal processing, machine learning, and data science be used for efficient data acquisition and inference? Justin Romberg is the Schlumberger Professor and the Associate Chair for Research in the School of Electrical and Computer Engineering. His research interests lie on the intersection of signal processing, machine learning, optimization, and applied probability. In the Fall of 2006, he joined the Georgia Tech ECE faculty. His work investigates structured signal models, compressed sensing, and computational methods for high-dimensional data analysis.
Lang Tong Research Group
Cornell UniversityLang Tong · Irwin and Joan Jacobs Professor of Engineering
How can optimization and machine learning improve the market operations of smart grid and future energy systems? Lang Tong joined Cornell University in 1998 and is the Cornell site director of the Power Systems Engineering Research Center. His current research focuses on optimization, machine learning, and market operations of smart grid and future energy systems. Research interests also include electric vehicles and smart transportation systems. Teaching interests include statistical inference, machine learning, optimization, power and energy systems, and electricity markets.
Lieven Vandenberghe Research Group
University of California, Los AngelesLieven Vandenberghe · Professor
How can optimization methods be used in engineering, signal processing, and machine learning? Lieven Vandenberghe lists research interests in engineering optimization, system theory and control, signal processing, and machine learning. His notable publications include the book Convex Optimization from Cambridge University Press, dated 2004. His publications also include a SIAM Review paper titled Semi definite programming. His department page also lists courses including applied numerical computing and a course on simulation, optimization, and data analysis.
Martin W. P. Savelsbergh Research Group
Georgia Institute of TechnologyMartin W. P. Savelsbergh · James C. Edenfield Chair and professor
How can optimization improve last-mile delivery and logistics? Martin Savelsbergh is the James C. Edenfield Chair and professor in the H. Milton Stewart School of Industrial and Systems Engineering. He is known for advances in dynamic ridesharing and methods for multi-objective optimization. He is also known for dynamic management of time-expanded networks. He has investigated with John-Paul Clarke how drones, in combination with ground-delivery vehicles, can be used for rapid home deliveries.
Martin Wainwright Research Group
Massachusetts Institute of TechnologyMartin Wainwright · Professor
How can statistics, machine learning, information theory and optimization be advanced together? Martin Wainwright is a professor in Electrical Engineering and Computer Science and Mathematics at MIT. He is affiliated with the Laboratory for Information and Decision Systems and the Statistics and Data Science Center. He joined the MIT faculty in July 2022 from the University of California at Berkeley. He has also authored a book on high dimensional statistics, and has co-authored books on graphical models and sparse statistical modeling.
Michael Frank Research Group
Brown UniversityMichael Frank · Edgar L. Marston Professor of Psychology
How do neural mechanisms support reinforcement learning, decision making and cognitive control? Michael Frank directs the Lab for Neural Computation and Cognition, which develops computational models at multiple levels of description, from neural circuit implementations to higher-level algorithmic and computational levels. He uses models of motivated behavior to quantify and explain variation due to genetics and neural activity. He also uses ML methods for interrogating neural data to test theories, helping to establish the nascent field of computational psychiatry. He and his students have created online repositories and tutorials for community software, including a hierarchical Bayesian parameter estimation toolbox for exploring quantitative relationships between neural and behavioral measures.
Nick Sahinidis Research Group
Georgia Institute of TechnologyNick Sahinidis · Gary C. Butler Family Chair and Professor
How can theory, algorithms, and software advance global optimization of mixed-integer nonlinear programs? Sahinidis conducts research at the interface of optimization, machine learning, and computational science. His contributions include mixed-integer nonlinear optimization, sparse model discovery, and data-driven optimization of complex systems. He has deployed his algorithms to solve problems in (bio)molecular engineering, energy systems, and supply chains. His research activities also include informatics problems in chemistry and biology and process and energy systems engineering.
Pedro Felipe Felzenszwalb Research Group
Brown UniversityPedro Felipe Felzenszwalb · Professor of Engineering and Computer Science
How can ideas from artificial intelligence and optimization help computers solve problems in computer vision? Pedro Felipe Felzenszwalb's main research interests are in computer vision, artificial intelligence, machine learning, optimization and signal processing. He has worked on a range of problems within computer vision, including the problem of image restoration. He has also worked on the problem of image segmentation and the problem of object recognition. His research connects computer vision and artificial intelligence to areas including optimization, statistics, and natural language processing.
Pierre Baldi Research Group
University of California, IrvinePierre Baldi · Distinguished Professor
How can deep learning and neural networks help explain natural and artificial intelligence? Pierre Baldi works on AI and machine learning, with emphasis on deep learning, neural networks, reinforcement learning, and their theoretical foundations and applications. Applications are pursued particularly in the natural sciences. In chemistry, applications include prediction of molecular properties, prediction of chemical reactions, drug discovery, and chemoinformatics. In biology, applications include neuroscience, circadian rhythms, gene regulation, omic sciences, protein structure prediction, bioinformatics, and systems biology.
Sergey Levine Research Group
University of California, BerkeleySergey Levine · Associate Professor
How can machines learn to make good decisions and control robots by combining perception with deep reinforcement learning? His work focuses on machine learning for decision making and control, with an emphasis on deep learning and reinforcement learning algorithms. His research includes developing algorithms for end-to-end training of deep neural network policies that combine perception and control. His research also includes scalable algorithms for inverse reinforcement learning, along with deep reinforcement learning algorithms. Applications of his work include autonomous robots and vehicles, as well as computer vision and graphics, and his listed research centers include the Berkeley Artificial Intelligence Research Lab and Berkeley Deep Drive.
Zhaojian Li Research Group
Michigan State UniversityZhaojian Li · Red Cedar Distinguished Professor, Mechanical Engineering
Can robots pick and sort apples reliably in unstructured orchard environments using integrated mobile manipulation systems? The group is the PI and lead developer for an Automated and Integrated Mobile System (AIMS) project to develop an efficient robotic apple harvester funded through a USDA Specialty Crop Research Initiative. The researchers design cloud-enabled vehicle control and sensing methods to address connectivity, control and privacy in connected and automated vehicles as described in the CAREER project. The team builds soft multi-arm robotic manipulators (SMART) and applies learning-based controllers to enable safer, compliant human-robot collaboration. The group applies learning and control of robot systems combined with vehicle dynamics and optimal control to improve autonomous mobility and robotic manipulation.
Showing 1–22 of 22 labs