PhD Students

BARREA Allan (Starting date : September 2013)
The role of finger pad mechanics in dexterous manipulation
Thesis advisors : Ph. Lefèvre, J-L Thonnard
This project focuses on the mechanical properties of the fingertip and its effects on the modulation of grasping force during object manipulation. We will use robotic devices and platforms in normal gravity as well as in microgravity to precisely assess the influence of constraints (rotational and/or tangential) on skin deformation and grip control. Moreover, we will carry out psychometric experiments to study how fingertip slips are perceived by human subjects.

CHEVALIER Pierre-Yves (Starting date : September 2012)
Decentralized computation and information fusion in a dynamical environment
Thesis advisors : J. Hendrickx, R. Jungers, R. Sepulchre (ULg)
We consider a set of agents having each a certain information or measure. The goal is to compute a function of all the available information in a decentralized way, giving a priori similar computing roles to all the agents.
We will consider this problem in the context of multiple dynamics: (i) The network of connections between the agents evolves, while possibly temporarily preserving certain local structure. (ii) The information accessible to the agents may change with time. (iii) The very composition of the system may also by dynamics, with agents entering and leaving the system.

CORDOVA BULENS David (Starting date : September 2013)
Dextrous Manipulation in Microgravity: Study of Motor Planning through Interaction and Perturbation with Robots
Thesis advisor : P. Lefèvre, J-L Thonnard
In most daily activities humans use both hands at the same time, whether it is to open a bottle or to transport large objects. In order for these activities to be efficient, coordination between hands is necessary. How this coordination impacts motor planning remains widely unknown and thus is an interesting domain to explore. We will also investigate the role gaze-grasping coordination and the influence of the alteration of certain senses through the use of virtual reality. Two tools are used in order to study these aspects : parabolic flights and robotic arms (the Phantom by SensAble and the Kinarm by b-kin).

DERAVET Nicolas (Starting date : September 2012)
Study of the oculomotor memory and application to the diagnosis of frontotemporal degeneration (FTLD)
Thesis advisor : P. Lefèvre
This thesis investigates the oculomotor memory, which plays a decisive role in the visual tracking system. The oculomotor memory takes part in the anticipation and the prediction of the displacement of moving objects in our surrounding environment. A recent study, performed in the laboratory of my promotor, has shown that a better understanding of this memory can be a very useful tool for the early diagnosis of some diseases. Indeed, the study of the eye movements of patients has allowed to clearly discriminate between patients with fronto-temporal lobar degeneration (FTLD) and those with Alzheimer disease (AD) or healthy subjects. During this thesis, I will study the oculomotor memory following several approaches, with healthy subjects and different categories of patients (FTLD, AD, MCI: Minor Cognitive Impairment). Finally, I will develop a specific toolkit, in two complementary stages. The first will aim at improving and optimizing our experimental setup to allow for faster and easier operation, whereas the second will aim at simplifying this setup to open new prospects of use of our tools in a "standard" clinical environment (able to measure the eye movements of patients). This interdisciplinary project combines an experimental approach based on fundamental research in neuroscience with a clinical study and the use of engineering tools.

DEWEZ Julien (Starting date : September 2013)
Exact Nonnegative Matrix Factorization: Algorithms, Bounds and Applications to Optimization
Thesis advisor : F. Glineur
In linear programming, when the number of constraints grows exponentially with the dimension, the time required to compute the solution also grows exponentially. One way to reformulate the problem with fewer constraints is to realize that the set of constraints defines a polytope with as many facets as there are constraints. We look for an extension of this polytope of feasible solutions with few facets such that solving the linear program on the extension reduces the computation time. Finding an extension for a convex polytope can be done with Exact Nonnegative Matrix Factorization, which is the central problem that we will study. The goals of this research are (i) the development of algorithms that compute exact nonnegative factorizations and (ii) the development of lower bounds (optimality guarantees for the algorithms) on the minimum inner dimension of such factorizations.

DI GIROLAMO Giovanni (Starting date : September 2015)
Optimization and control over wireless networks
Thesis advisor : R. Jungers
More information soon.

DONG Shuyu (Starting date : June 2016)
Thesis advisor : P.-A. Absil
More information soon.

GENICOT Matthieu (Starting date : September 2014)
Matrix factorization techniques applied to fMRI-EEG coupled datasets
Thesis advisors : P.-A. Absil, R. Lambiotte (UNamur)
This research project aims to develop algorithms based on nonnegative matrix factorization (NMF) to analyze bi-modal datasets composed of electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) data. In this work, different versions of NMF (i.e. with different constraints) adapted to multi-modal datasets will be implemented and tested on EEG-fMRI datasets. Such adaptation of matrix factorization technique to multi-modal datasets has already been performed for ICA (independent component analysis), giving rise to joint ICA, which has been successfully used in the context of EEG-fMRI coupled data. It is expected that the NMF techniques will provide an easier interpretation of the data in view of its 'part-based' representation property.

GIARD Thibault (Starting date : September 2010)
Dextrous Manipulation in Microgravity: Motor Planning and Interaction with Robots
Thesis advisors : P. Lefèvre, J-L Thonnard
With the thumb opposing the index, humans have the ability to grasp and manipulate objects. More specifically, this grip permits to modulate the force developed to hold objects stable in our hands. Indeed, the force required to hold an object depends on many factors, including the object weight, the mechanical and frictional properties of the skin/object interface and the inertial constraints induced when the object is accelerated. In most situations encountered on daily bases, manipulating an object also implies the anticipation of static and inertial torques. In this project, two tools are used in order to study those aspects : parabolic flights and robotic arms (the Phantom by SensAble and the Kinarm by b-kin).

GONZE François (Starting date : February 2014)
Synchronizing automatons
Thesis advisor : R. Jungers
The Czerny conjecture is the most longstanding unproved conjecture in the automata theory. We are using the concept of synchronising probability function of an automaton to approach this problem. That way, we can apply linear programing techniques that have not been used in the past to understand synchronising automatons, with the aim to prove the conjecture, or to find some counter examples to it.

GOUSENBOURGER Pierre-Yves (Starting date : September 2014)
Interpolation on Riemannian manifolds of high dimension
Thesis advisors : P.-A. Absil, L. Jacques
It is now well established that manifold modeling can be applied to many different fields including data mining, image modeling and processing, invariant pattern recognition, and interpolation and optimization on manifolds. The aim of this project is to advance the state of the art both in interpolation and approximation on manifolds, with applications in particular to data restoration and compression.

GUEUNING Martin (Starting date : September 2013)
Methods and Models for Temporal Networks
Thesis advisors : J-C Delvenne, R. Lambiotte (UNamur)
Our goal is to develop a theoretical framework for dynamical, or temporal, networks taking temporal patterns into account. We are also interested in developping efficient algorithms extracting and exploiting information in temporal and relational data (collected on real-life networks), including community detection and ranking.

GUTIERREZ GOMEZ Leonardo (Starting date : September 2015)
Mining of large temporal networks
Thesis advisor : J.-C. Delvenne
It is customary to characterize Big Datasets by their Volume, Velocity, and Variety. While posing challenges on their own, these characteristics also offer unprecedented opportunities. In this research, we develop data mining methods adapted to mining of time-stamped data in order to exploit the new possibilities offered by these extra dimensions. Concretely, we use deep learning methods for discovering communities and predicting time-varying evolution on networks.

LATIERS Arnaud (Starting date : October 2012)
Dynamic demand response as provider of ancillary services
Thesis advisors : F. Glineur, E. De Jaeger
Describe, model and understand dynamic demand response and assess its potential as a provider of ancillary services (flexibility) in a high-voltage grid.

LEGAT Benoît (Starting date : October 2016)
Thesis advisor : R. Jungers
More information soon.

MASSART Estelle (Starting date : September 2015)
Data fitting on manifolds
Thesis advisors : P.-A. Absil, J. Hendrickx
Our goal is to develop the required theory and to implement algorithmic solutions for three data-fitting problems on manifolds (barycenter computation, approximation of a set of data points by a curve, and learning of a mapping between two manifolds), with applications notably in computer vision and image analysis.​

PHILIPPE Matthew (Starting date : September 2013)
Decentralized control of hybrid systems
Thesis advisor : R. Jungers
Applications requiring, for practical and/or technological reasons, decentralized control are plentiful (smart grids, internet's congestion control ...). Often, the agents participating in the control are of hybrid nature: to completely describe their behavior, one must study dynamical systems that may “switch” whenever some conditions are met. Our goal in this thesis is to see what can be achieved by actively taking into account the hybrid nature of agents operating in a distributed system. Our main inspiration for the project is internet's congestion control, and we plan to apply our knowledge to this domain

RENARD Emilie (Starting date : January 2013)
Dimensionality reduction using matrix factorization
Thesis advisors : P-A Absil, V. Blondel
Datasets with high-dimensionnality are more and more common today and necessitate adapted methods. The goal here is to develop methods to factorize "large p small n" matrices into a product of two matrices of smaller rank k. A classic example are microarray data of which dimensions are typically a few thousands of features for only a few hundreds of observations. Different types of constraints can be considered: sparsity, non-negativity, statistical independence,... and ideally the methods should be stable with respect to k, and the number of observations.

ROMO HERNANDEZ Aarón (Starting date : September 2015)
Thermodynamic control
Thesis advisor : D. Dochain
Thermodynamics is one of the most powerful and most elegant of the engineering disciplines. It can be understood as one of the sciences that lies behind energy-dynamics. Lyapunov theory sets particular properties for non-linear dynamical systems. Thermodynamic functions and Lyapunov analysis permit to link with ease: the stability of a process evolution and the natural phenomena occurring inside it. This connection is well developed for mechanical and electrical structures, mostly. In chemical processes, however, we have renounced to this phenomenological backup in the pursuit of purely abstract analysis.
In this work we will set a methodology that prosecutes phenomenological analysis in chemical systems. We will go after this approach using thermodynamic functions as a departure point. Irreversible thermodynamics is preferred since it naturally defines functions that appear as Lyapunov criteria candidates. Flash drum dynamics will be studied as a pragmatic application of the methodology.

SALMEN Florian (Starting date : October 2013)
Perception and Action in Complex Dynamic Visual Environments
Thesis advisor : P. Lefèvre
Human beings have the remarkable possibility to predict the course of moving objects even after directional changes induced by collisions. This ability helps to align the movement of hands and arms to succeed in a multiplicity of situations like sports or in simple everyday life tasks like catching thrown down objects. Until now it is not completely understood how this predictive ability works. Therefore we want to investigate the mechanisms behind it and study how or if the causal perception is connected with the oculomotor response of the eye. Since object collisions in real life do not take place under simple or reduced conditions most of the time, we consider perception and action in complex dynamic visual environments.

TAYLOR Adrien (Starting date : September 2012)
Control and optimization of networked systems with poorly understood couplings; with application to wind farms
Thesis advisors : J. Hendrickx, F. Glineur
Classical control and optimization techniques rely either on the fact we know how the studied system behaves (i.e. we have a reliable model of it) or on black-box models. Our aim is to develop methods standing in-between those two approaches: using partial informations about the system (e.g. bounds on the function or the derivative). This is in particular of interest in the case of wind farms, where the interactions between the different wind turbines - through wind field - cannot be computed accurately in a reasonable time.

TREFOIS Maguy (Starting date : September 2010)
Algorithmics of structured matrices and algebraic graph theory
Thesis advisor : J.-C. Delvenne
We investigate the minimum rank problem of a graph, that is the minimum possible rank of a real matrix whose zero-nonzero pattern is described by the graph. The project consists in developing efficient methods allowing to compute the minimum rank of particular graphs, namely the loop directed trees. Moreover, we work on complex networks and more precisely we develop methods in order to reduce the size of a network by preserving its dynamics.

VANDE KERCHOVE Corentin (Starting date : September 2013)
Social behaviour analysis through interactions using consensus models
Thesis advisor : V. Blondel
The purpose of the research is to provide well-defined characteristics as well as efficiency factors for a group, in order to reach an optimal consensus. Ultimately, it will focus on drawing a model which will allow to predict the group's efficiency, and explain the resulting decisions by the human personality traits. Choosing meaningful features involves the necessity to extract an underlying dimension of the opinion dynamics, i.e. identify principal components in order to characterize the social interactions.

VINCENT Benjamin (Starting date : September 2015)
Inferential control of structured distributed thermodynamics systems. Application to tokamak
Thesis advisor : D. Dochain
We will study and develop control strategies based on state observers and parameter estimators for irreversible distributed thermodynamic systems. The innovation in this work resides in the use of structured systems, such as Port-Hamiltonian and metriplectic formulations. The developed control strategies will be applied to the control of plasma profiles in tokamaks.

VITAL JACOME Miguel Angel (Starting date : August 2015)
Applications of microrespirometry for characterization of bioprocess
Thesis advisor : D. Dochain
My research is focused on the problem of parameter calibration of mathematical models used in bioprocess. We are developing novel techniques based in microrespirometry (respirometry in microreactors), which allows to obtain high quality and quantity data, compared with traditional techniques. Our goal is to study the identifiability problems and determine the uncertainty of kinetic parameters of selected models, as well as optimizing data analysis, in order to reach the maximum potential of these techniques. Some applications are being developed to characterize the effect of temperature in microbial kinetics and to study aerobic granular biofilms.

XIANCHAO Tang (Starting date : September 2015)
Clustering methods for social networks
Thesis advisor : J.-D. Delvenne
More information soon.

Post-Docs and Research Associates

ATHALYE Sanand (Starting date : October 2015)
Feedback stabilization with decentralized control and network coding
Host : R. Jungers
We want to stabilize a linear system by state/output feedback when controllers and plant are geographically distributed. (In literature, problem with one centralized controller is already considered.) We quantize state/output information and transmit it via communication channels to geographically distributed controllers. There are bandwidth constraints on the channels. We look for conditions to stabilize the system (by optimizing bandwidth constraints on the channels) using network coding.

ATHANASOPOULOS Nikolaos (Starting date : March 2015)
Set-theoretic methods for analysis and control of dynamical systems
Host : R. Jungers
More information soon.

CREVECOEUR Frédéric (Starting date : October 2014)
Arm-hand Coordination for Feedback Control during Object Manipulation
Host : P. Lefèvre
More information soon.

EGO Caroline (Starting date : July 2015)
Visual tracking in cerebral-palsied children: influence on learning and exploration of new avenues for rehabilitation
Host : Ph. Lefèvre
More information soon.

FACCIN Mauro (Starting date : July 2015)
Dynamics of and on graphs, clustering, non-Markovian systems
Host : J.-C. Delvenne
Network theory is a powerful tool that is applied in a wide range of research domains to a everyday increasing number of problems. In this framework, diffusion processes are used to model a great variety of systems. My research focus on understanding and characterize diffusion processes on static and dynamic networks, including processes with memory that break the Markov property. This properties can be used as a probe to unveil mesoscopic properties of the underlying network, such as communities or modules in general.

GUSEV Vladimir (Starting date : April 2015)
Synchronizing automata
Host : R. Jungers
More information soon.

HUANG Wen (Starting date : September 2014)
Riemannian optimization: theory and applications
Host : P.-A. Absil
Riemannian optimization, also called optimization on manifolds or differential-geometric optimization, concerns finding an optimum (global, or more reasonably, local) of a real-valued function defined over a nonlinear search space that admits the structure of a (smooth) manifold. By definition, manifolds are sets covered with coordinate patches that overlap smoothly. They generalize the familiar smooth curves and surfaces. Riemannian optimization is motivated by various problems in the sciences and engineering that can be formulated in this framework. We propose to advance the state of the art in Riemannian optimization along two broad directions. (i) The design, analysis, and implementation of general-purpose optimization methods on Riemannian manifolds. (ii) The application of such algorithms to various computational problems, notably those featuring low-rank matrix constraints, where there is strong evidence that the manifold-based methods are capable of outperforming existing approaches.

MOENS Luc (Starting date : October 1994)
Identification of a model for the real-time application for riverflow forecasting
HYDROMAX is a real-time application for riverflow forecasting which is developed by CESAME and is operational at SETHY (Service d'études hydrologiques, Walloon Ministry of Public Works - Belgium) for the management of flood alarms and the on-line information of the rescue services in the Meuse river basin. HYDROMAX is connected with the telemetering network of SETHY. This network is essentially made up of water level chart recorders and rain gauges.
HYDROMAX provides in real-time:

  • short term predictions of riverflows based on rainfal and past riverflow measurements
  • long term flood forecastings based on weather forecasts

For each river basin, the predictions are produced by a mathematical model. HYDROMAX has been developed to be user friendly and to fulfill the real time forecasting requirements. HYDROMAX is successfully in operation since 1995 on the main Meuse tributaries (Semois, Ourthe, Lesse, Viroin, etc.).

PEEL Leto (Starting date : August 2015)
Statistical inference in complex networks
Host : Jean-Charles Delvenne
More information soon.

SCHAUB Michael (Starting date : December 2014)
Complex networks and dynamics
Host : Jean-Charles Delvenne
My research deals with the analysis of complex systems that can be abstracted as networks or graphs. Using tools from graph theory and dynamical systems I am trying to dynamically reveal relevant structure (e.g. community structure) in these systems. Applications include bio-chemical systems but also social and technological networks. Apart from the analysis of networks in terms of their community structure I am broadly interested in the analysis and modelling of network like systems. In particular I am interested in the relation between structure, dynamics and function; information transfer and retrieval (especially in neural systems); information diffusion in networks; and control of networked systems.

SIMONETTO Andrea (Starting date : March 2016)
Distributed and Large-Scale Optimization with Air traffic control applications
Host : Raphaël Jungers
My research focuses on distributed convex optimization; I have made contributions in different application fields, namely control, estimation, and signal processing, as well as in more theoretical areas in convex optimization. At UCL, I will work on optimization algorithms capable of handling large-scale and networked systems, with a particular emphasis on air traffic control applications. My work is sponsored by the H2020 COPTRA project and involves a good deal of optimization, control, machine learning, and big data analytics.

STICH Sebastian (Starting date : November 2014)
Mining and optimization of big data models
Host : François Glineur
More information soon.

Useful links : UCL | KUL | UGent | VUB | ULg | UMons | UNamur | Stanford | Princeton | MIT