Kasteelpark Arenberg 10

Heverlee, Belgium

Kasteelpark Arenberg 10

Heverlee, Belgium
Time filter
Source Type

Renkens V.,Kasteelpark Arenberg 10 | Van hamme H.,Kasteelpark Arenberg 10
Signal Processing | Year: 2017

In this paper a method for model order selection through automatic relevance determination in NMF is proposed. Overfitting is avoided by inferring the relevance of components in the dictionary and removing the irrelevant ones. To reduce the number of parameters in the model the activations are treated as missing data and marginalized out, the marginal posterior is maximized. Furthermore the hyper-parameters are found with full Bayesian inference, so they no longer have to be tuned. The proposed method solves two suboptimalities in a previously proposed method for automatic relevance determination based on the maximization of the joint posterior. (1) In the joint posterior method the activations of the dictionary components are inferred by training them as parameters of the model and (2) the hyper-parameters are chosen by hand and have to be tuned. The proposed algorithm is extensively tested on a synthetic dataset and the swimmer dataset. Results from a face reconstruction task on the CBCL dataset and an unsupervised spoken keyword discovery task on the TIDIGITS dataset are also presented. The results show that the proposed algorithm outperforms previously proposed algorithms in most experiments. © 2016

Welkenhuysen M.,Catholic University of Leuven | Welkenhuysen M.,IMEC | Gligorijevic I.,Kasteelpark Arenberg 10 | Gligorijevic I.,Future Health | And 5 more authors.
Behavioural Brain Research | Year: 2013

In search of a new potential target for deep brain stimulation in patients with obsessive-compulsive disorder (OCD), we evaluated the single-cell activity of neurons in the bed nucleus of the stria terminalis (BST) in urethane-anesthetized rats in an animal model for OCD, the schedule-induced polydipsia (SIP) model, and compared this to the BST activity in control rats and to a third group of rats which were introduced in the model but did not develop the SIP, and thus were considered resistant. We compared the firing rate and firing pattern of BST neurons between these groups, between hemispheres and made a correlation of the firing rate and firing pattern to the position in the BST. The variability of BST neurons in SIP rats was lower and the randomness higher than BST neurons in control rats or resistant rats. The firing rate of BST neurons in SIP rats was significantly higher and the burst index lower than BST neurons in resistant rats but not in control rats. Also, neurons from the right hemisphere in the SIP group had a higher burst index than neurons from the left hemisphere. However, this is opposite in the resistant and control group. Third, we found a higher bursting index with increasing (more ventral) depth of recording. These findings suggest that schedule-induced polydipsia, which models compulsive behavior in humans, induces a change in firing behavior of BST neurons. © 2012 Elsevier B.V.

Seslija M.,Kasteelpark Arenberg 10 | Van Der Schaft A.,University of Groningen | Scherpen J.M.A.,University of Groningen
Automatica | Year: 2014

Inspired by the recent developments in modeling and analysis of reaction networks, we provide a geometric formulation of the reversible reaction networks under the influence of diffusion. Using the graph knowledge of the underlying reaction network, the obtained reaction-diffusion system is a distributed-parameter port-Hamiltonian system on a compact spatial domain. Motivated by the need for computer-based design, we offer a spatially consistent discretization of the PDE system and, in a systematic manner, recover a compartmental ODE model on a simplicial triangulation of the spatial domain. Exploring the properties of a balanced weighted Laplacian matrix of the reaction network and the Laplacian of the simplicial complex, we characterize the space of equilibrium points and provide a simple stability analysis on the state space modulo the space of equilibrium points. The paper rules out the possibility of the persistence of spatial patterns for the compartmental balanced reaction-diffusion networks. © 2014 Elsevier Ltd. All rights reserved.

Seslija M.,Kasteelpark Arenberg 10 | Scherpen J.M.A.,University of Groningen | Van Der Schaft A.,University of Groningen
Automatica | Year: 2014

Simplicial Dirac structures as finite analogues of the canonical Stokes-Dirac structure, capturing the topological laws of the system, are defined on simplicial manifolds in terms of primal and dual cochains related by the coboundary operators. These finite-dimensional Dirac structures offer a framework for the formulation of standard input-output finite-dimensional port-Hamiltonian systems that emulate the behavior of distributed-parameter port-Hamiltonian systems. This paper elaborates on the matrix representations of simplicial Dirac structures and the resulting port-Hamiltonian systems on simplicial manifolds. Employing these representations, we consider the existence of structural invariants and demonstrate how they pertain to the energy shaping of port-Hamiltonian systems on simplicial manifolds. © 2013 Elsevier Ltd. All rights reserved.

Tavernier F.,Kasteelpark Arenberg 10 | Steyaert M.,Kasteelpark Arenberg 10
Analog Integrated Circuits and Signal Processing | Year: 2011

A new TIA topology with enhanced bandwidth is presented in this paper. By adding an extra capacitive feedback loop to the resistive feedback TIA, bandwidth and sensitivity are increased without sacrificing the low power consumption. It is shown that this topology is superior to the self-compensated TIA when the photodiode is integrated on the same die as the TIA. An implementation is presented that boosts the bandwidth by a factor of 9 and reduces the noise by a factor of 4.2 for a photodiode capacitance of 106 pF, the parasitic capacitance of a POF-compliant 1 mm integrated photodiode in 130 nm CMOS. © 2010 Springer Science+Business Media, LLC.

Van Breussegem T.,Kasteelpark Arenberg 10 | Steyaert M.,Kasteelpark Arenberg 10
Analog Integrated Circuits and Signal Processing | Year: 2012

Contemporary models fail to include the influence of the output buffer capacitor size on the performance of capacitive DC-DC converters. This letter examines the relevance of this dependency and shows how to adapt existing models in order to include it. The improved model is verified mathematically for down-converters, by means of Spice simulations and based on measurements of silicon integrated prototypes. Measurements demonstrate an accuracy improvement of up to 30 % compared with the conventional model. © 2012 Springer Science+Business Media, LLC.

Callemeyn P.,Kasteelpark Arenberg 10 | Jonghe D.D.,Kasteelpark Arenberg 10 | Gielen G.G.E.,Kasteelpark Arenberg 10 | Steyaert M.S.J.,Kasteelpark Arenberg 10
Analog Integrated Circuits and Signal Processing | Year: 2014

The evolution of computer-aided design tools has extended the capabilities of a designer by pushing the optimality of complex circuits beyond the ad hoc manual implementation. This work presents a framework to co-optimize the circuit and the layout parameters of fully integrated inductive DC-DC converters. The framework comprises expensive optimization that is speeded up by active learning sample selection and evolutionary techniques to acquire an optimal converter. A tapered inductor topology is used to increase the quality of the on-chip inductor and to improve the efficiency of the overall monolithic DC-DC converter. The optimization framework is validated by co-optimizing the design parameters and the tapered inductor layout for a fully-integrated DC-DC boost converter in a 0.13 μm CMOS technology. The power loss in the circuit is reduced with 27 % resulting in a 7 % efficiency improvement, compared to a fully-integrated DC-DC boost converter with a regular inductor topology. © 2013 Springer Science+Business Media New York.

Langone R.,Kasteelpark Arenberg 10 | Suykens J.A.K.,Kasteelpark Arenberg 10
Journal of Physics: Conference Series | Year: 2013

This work is related to the problem of community detection in dynamic scenarios, which for instance arises in the segmentation of moving objects, clustering of telephone traffic data, time-series micro-array data etc. A desirable feature of a clustering model which has to capture the evolution of communities over time is the temporal smoothness between clusters in successive time-steps. In this way the model is able to track the long-term trend and in the same time it smooths out short-term variation due to noise. We use the Kernel Spectral Clustering with Memory effect (MKSC) which allows to predict cluster memberships of new nodes via out-of-sample extension and has a proper model selection scheme. It is based on a constrained optimization formulation typical of Least Squares Support Vector Machines (LS-SVM), where the objective function is designed to explicitly incorporate temporal smoothness as a valid prior knowledge. The latter, in fact, allows the model to cluster the current data well and to be consistent with the recent history. Here we propose a generalization of the MKSC model with an arbitrary memory, not only one time-step in the past. The experiments conducted on toy problems confirm our expectations: the more memory we add to the model, the smoother over time are the clustering results. We also compare with the Evolutionary Spectral Clustering (ESC) algorithm which is a state-of-the art method, and we obtain comparable or better results. © Published under licence by IOP Publishing Ltd.

Loading Kasteelpark Arenberg 10 collaborators
Loading Kasteelpark Arenberg 10 collaborators