Projects
Resource-Constrained Machine Learning across the Computing Continuum KU Leuven
Sensor networks typically gather information and centralize in a backend server application. However, considering the current evolution of the number of Internet of Things (IoT) devices this will amount to a huge network traffic and a substantial load of the backend application. This PhD topic wants to utilize the whole IoT infrastructure to distribute data processing steps. In this way, a multi-tier data processing chain can be constructed ...
Industry 4.0 solutions for optimised value recovery from Waste Electrical and Electric Equipment (WEEE) KU Leuven
Demanufacturing activities consist of the disassembly of products to their separate components for recovery or further recycling as well as the disassembly or destructive processing of the separate components for the recovery of valuable materials. Remanufacturing activities consist of the subsequent cleaning, repairing, as well as refurbishing, upgrading and maintenance of recovered components or entire product assemblies. The potential of ...
Distributed Network of Active Noise Equalizers for Multi-User Sound Control KU Leuven
The project is aimed at developing a novel wireless acoustic sensor network (WASN) for multi-user active noise equalization without the need of using headsets. This system is formed by multiple active noise equalizers (ANEs), which act as a node in a WASN-type set up and cooperate to simultaneously solve their node-specific noise equalization problems. By doing so, the proposed system can simultaneously ensure the auditory comfort of multiple ...
Optical interferometer for 3D surface characterisation with sub-nanometer depth resolution KU Leuven
Numerical methods for tensor decompositions. From least squares to beta-divergence, from batch to updating. KU Leuven
Over the years, many algorithms have been designed to decompose a matrix into a product of other matrices. These matrix decompositions can be used to compress data with a minimal loss of information or for extracting meaningful components. More recently, tensor decompositions such as the canonical
polyadic decomposition (CPD) and the low multilinear rank approximation (LMLRA) have been designed as higher-order generalizations of these ...
Newton-type operator splitting methods for real-time optimization of cyberphysical systems KU Leuven
Operator splitting techniques, introduced in the 50's for solving PDEs and optimal control problems, have been successfully used to reduce complex problems into a series of simpler subproblems. They have recently received an enormous renewed interest due to their ability in handling large-scale and embedded convex optimization problems, and thus have found numerous applications in real-time control, machine learning, data mining and signal ...
Supra-Threshold Electrically Evoked Auditory Steady-State Responses In Cochlear Implant Users KU Leuven
Cochlear implants (CIs) restore audition to the profoundly deaf by directly stimulating the auditory nerve. For accurate speech perception, incoming sound must be encoded in a meaningful way, and transmitted along the auditory pathway. CIs encode speech by modulating the amplitude of pulse train sequences, this thesis investigates neural responses to amplitude modulated stimuli in the CI population.
Envelope coding strategies facilitate ...