Projects
Structured low-rank matrix / tensor approximation: numerical optimization-based algorithms and applications KU Leuven
Today's information society is centered on the collection of large amounts of data, from which countless applications aim at extracting information. They involve the manipulation of matrices and higher-order tensors, which can be viewed as large multi-way arrays containing numerical data. Key to their successful and efficient processing is the proper exploitation of available structure, and in particular low rank. This project aims to ...
Numerical optimization algorithms for large-scale problems in machine learning and control KU Leuven
The topic of this PhD is to develop novel numerical optimization algorithms for solving large-scale problems arising in machine learning and control. The focus of the thesis will be threefold: 1. Theory: global convergence analysis, asymptotic rate analysis, saddle point avoidance for nonconvex problems, 2. Implementation: efficient numerical linear algebra routines, open-source software, 3. Applications: deep restricted kernel machines, ...
Numerical algorithms for large scale matrices with uncertain coefficients. KU Leuven
Numerical Tensor Algorithms KU Leuven
'Dynamic Spectrum Management' techniques and numerical optimization algorithms design for green, stable and fast DSL broadband access networks. KU Leuven
Numerical algorithms for tensor decompositions and applications. KU Leuven
The basic outline of the project is to develop new numerical algorithms for tensor decompositions developing connections with approximate invariant subspaces. These are tailored to applications in signal processing, such as blind source separation and latent variable analysis.
Advanced algoritms for optimal numerical integration. KU Leuven
Computational multiscale methods: algorithms, analysis and applications KU Leuven
Nowadays, computer simulation is essential to create insight in many scientific domains, due to the availability of advanced numerical methods. However, results from simulation are only as meaningful as the model that was used to describe the system. Many physical, chemical and biological problems have a multiscale nature, and cannot be modeled to sufficient accuracy at the macroscopic scale of interest. At the same time, direct simulation ...
Pole swapping methods for the eigenvalue problem - Rational QR algorithms KU Leuven
The matrix eigenvalue problem is often encountered in scientific computing
applications. Although it has an uncomplicated problem formulation, the best
numerical algorithms devised to solve it are far from obvious.
Computing all eigenvalues of a small to medium-sized matrix is nowadays a
routine task for an algorithm of implicit QR-type using a bulge chasing technique.
On the other hand projection methods are ...