About

Mohamed Ali Belabbas is a professor at The Grainger College of Engineering at the University of Illinois Urbana-Champaign. His research focuses on nonlinear systems and control, with specific interests in artificial intelligence and autonomous systems, including autonomous vehicular technology and UAVs. His work also encompasses bioelectronics, bioinformatics, cognitive computing, computational science and engineering, cyberphysical systems, data science, machine learning, network science, neuro-engineering, robotics, sensing systems, and smart infrastructures. He has taught courses such as Introduction to Robotics, Principles of Safe Autonomy, Control Systems, Control of Stochastic Systems, and Geometric Control Theory. Belabbas has received recognition for his contributions, including an NSF CAREER Award, and is involved in designing technology with security considerations in mind. His research aims to advance understanding and development in autonomous systems, control theory, and related fields, contributing to the broader engineering and scientific community.

Research topics

• Artificial Intelligence
• Computer Science
• Physics
• Mathematics
• Algorithm
• Engineering
• Mathematical physics
• Pure mathematics
• Classical mechanics
• Applied mathematics

Selected publications

• Constructing Stochastic Matrices for Weighted Averaging in Gossip Networks

  IFAC-PapersOnLine · 2025-01-01 · 1 citations

  articleOpen accessSenior authorCorresponding

  The convergence of the gossip process has been extensively studied; however, algorithms that generate a set of stochastic matrices, the infinite product of which converges to a rank-one matrix determined by a given weight vector, have been less explored. In this work, we propose an algorithm for constructing (local) stochastic matrices based on a given gossip network topology and a set of weights for averaging across different consensus clusters, ensuring that the gossip process converges to a finite limit set.

  PublisherDOI

• Control Theoretic Approach to Fine-Tuning and Transfer Learning

  Lecture notes in control and information sciences - proceedings · 2025-01-01

  book-chapter
  PublisherDOI

• Age of Coded Updates in Gossip Networks Under Memory and Memoryless Schemes

  IEEE Transactions on Communications · 2025-08-01 · 1 citations

  article

  We consider an information update system on a gossip network, where a source node encodes information into <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">n</i> total keys such that any subset of at least <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">k</i> + 1 keys can fully reconstruct the original information. This encoding process follows the principles of a <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">k</i>-out-of-<italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">n</i> threshold system. The encoded updates are then disseminated across the network through peer-to-peer communication. We have two different types of nodes in a network: subscriber nodes, which receive a unique key from the source node for every status update instantaneously, and nonsubscriber nodes, which receive a unique key for an update only if the node is selected by the source, and this selection is renewed for each update. For the message structure between nodes, we consider two different schemes: a memory scheme (in which the nodes keep the source’s current and previous encrypted messages) and a memoryless scheme (in which the nodes are allowed to only keep the source’s current message). We measure the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">timeliness</i> of information updates by using a recent performance metric called, the version age of information. We present explicit formulas for the time average AoI in a scalable homogeneous network as functions of the number of subscriber nodes under a memoryless scheme. Additionally, we provide strict lower and upper bounds for the time average AoI under a memory scheme.

  PublisherDOI

• Super-Linearization with Monomial Observables: Necessary and Sufficient Conditions

  IFAC-PapersOnLine · 2025-01-01

  articleOpen accessSenior authorCorresponding

  We establish necessary and sufficient conditions for polynomial systems with a unique visible monomial observable to be (strongly) super-linearized using monomial observables. Our analysis reveals that, while a super-linearization may include additional observables, the observables that are essential for achieving the super-linearization must all have the same degree as the visible observable.

  PublisherDOI

• Structural System Theory over Random Networks

  2025-12-09

  article1st authorCorresponding

  In this tutorial, we will introduce notions of structural systems theory, namely structural controllability and structural stability, and show how to study them over random network models, namely Erdős-Rényi random graphs and graphons. We give some proofs when they illustrate some universal techniques.

  PublisherDOI

• On Infinite-horizon Minimum Energy Control

  ArXiv.org · 2025-02-25

  preprintOpen access1st authorCorresponding

  We address the infinite-horizon minimum energy control problem for linear time-invariant finite-dimensional systems $(A, B)$. We show that the problem admits a solution if and only if $(A, B)$ is stabilizable and $A$ does not have imaginary eigenvalues.

  PublisherOA PDFDOI

• Constructing Stochastic Matrices for Weighted Averaging in Gossip Networks

  ArXiv.org · 2025-02-27

  preprintOpen accessSenior author

  The convergence of the gossip process has been extensively studied; however, algorithms that generate a set of stochastic matrices, the infinite product of which converges to a rank-one matrix determined by a given weight vector, have been less explored. In this work, we propose an algorithm for constructing (local) stochastic matrices based on a given gossip network topology and a set of weights for averaging across different consensus clusters, ensuring that the gossip process converges to a finite limit set.

  PublisherOA PDFDOI

• Interpretable Gradient Descent for Kalman Gain

  ArXiv.org · 2025-07-18

  preprintOpen access1st authorCorresponding

  We derive a decomposition for the gradient of the innovation loss with respect to the filter gain in a linear time-invariant system, decomposing as a product of an observability Gramian and a term quantifying the ``non-orthogonality" between the estimation error and the innovation. We leverage this decomposition to give a convergence proof of gradient descent to the optimal Kalman gain, specifically identifying how recovery of the Kalman gain depends on a non-standard observability condition, and obtaining an interpretable geometric convergence rate.

  PublisherOA PDFDOI

• Geometric Foundations of Tuning without Forgetting in Neural ODEs

  ArXiv.org · 2025-09-03

  preprintOpen access

  In our earlier work, we introduced the principle of Tuning without Forgetting (TwF) for sequential training of neural ODEs, where training samples are added iteratively and parameters are updated within the subspace of control functions that preserves the end-point mapping at previously learned samples on the manifold of output labels in the first-order approximation sense. In this letter, we prove that this parameter subspace forms a Banach submanifold of finite codimension under nonsingular controls, and we characterize its tangent space. This reveals that TwF corresponds to a continuation/deformation of the control function along the tangent space of this Banach submanifold, providing a theoretical foundation for its mapping-preserving (not forgetting) during the sequential training exactly, beyond first-order approximation.

  PublisherOA PDFDOI

• Time-delay Induced Stochastic Optimization and Extremum Seeking

  2025-06-24

  article

  In this paper a novel stochastic optimization and extremum seeking algorithm is presented, one which is based on time-delayed random perturbations and step size adaptation. For the case of a one-dimensional quadratic unconstrained optimization problem, global exponential convergence in expectation and global exponential practical convergence of the variance of the trajectories are proven. The theoretical results are complemented by numerical simulations for one-and multidimensional quadratic and non-quadratic objective functions.

  PublisherDOI

Recent grants

• CAREER: Graphs, geometry and algorithms for sparse decentralized control systems

  NSF · $400k · 2014–2020

• Multi-agent systems with localized objectives

  NSF · $291k · 2013–2017

• NSF/ENG/ECCS-BSF: Collaborative Research: Foundations of secure multi-agent networked systems

  NSF · $220k · 2018–2022

Frequent coauthors

• Xudong Chen

  57 shared

• Tamer Başar

  34 shared

• Shenyu Liu

  Beijing Institute of Technology

  11 shared

• Yinai Fan

  7 shared

• Patrick J. Wolfe

  Purdue University West Lafayette

  6 shared

• Ji Liu

  6 shared

• Artur Kirkoryan

  5 shared

• Brian D. O. Anderson

  Australian National University

  4 shared

Labs

• The Grainger College of EngineeringPI

Education

• Ph.D., Electrical Engineering

  University of California, Berkeley

  1991

• M.S., Electrical Engineering

  University of California, Berkeley

  1987

• B.S., Electrical Engineering

  University of Texas at Austin

  1985

Awards & honors

• NSF CAREER Award