All Years Seminars
[INMA] 2020-09-29 (14h) : MAYO: a pipeline for planetary systems imaging
At Euler building (room A.002) + MS Teams
Speaker :
Pairet, Benoît
Abstract :
Direct imaging of planetary systems is a challenging task mainly because of the tiny angular separation between faint circumstellar signal and the overwhelming bright star. Even with state-of-the art instruments this task heavily relies on post-processing. Indeed, the acquired images are plagued with bright quasi-static spots, called speckles, that hide the astrophysical signal, made of exoplanets and circumstellar disks. The latter are extended structure, which are inherently difficult to image and methods tailored for exoplanets imaging highly distort the morphology of disks. We thus propose a processing pipeline based on prior knowledge about disks, exoplanets and speckles that produces faithful images of disks, both in terms of morphology and intensity distribution. In addition, our pipeline is designed to separate exoplanets from disks.
[INMA] 2020-09-22 (14h) : Mathematics in medical physics for radiation oncology
At Euler building (room A.002) + MS Teams
Speaker :
Lee, John Aldo
Abstract :
Radiation oncology treats cancer by irradiating patients locally with ionizing beams, like X-rays, protons or even heavier particles.
Treatment planning can be seen as a ballistic problem whose solution is such that beams reach a certain dose on a target volume to control the tumor, while they do not exceed some threshold dose in organs at risk to limit adverse effects.
In other words, treatment planning can be translated into a multi-objective optimization problem or a constrained optimization problem, the former allowing better "tweakability" to satisfy complex clinical objectives.
Current research in radiation oncology, including in the UCLouvain/MIRO group, aims at modeling treatment uncertainties with multiple error scenarios in minimax problems, to make treatment more robust against those.
Other lines of investigation go towards faster, more user-friendly multi-objective interaction with Pareto frontier pre-calculation.
Yet another direction aims at improving accuracy in objective function evaluation by replacing fast "analytical" dose computation with Monte Carlo simulations.
Depending on simulation speed, hybrid optimization schemes can be devised, interleaving slow, accurate checkpoints of the objective function with faster, cheaper surrogates.
Further perspectives include clinical decision support system involving artificial intelligence.
This entails automating the few last manual steps in treatment planning, simulation, and evaluation with respect to clinical endpoints.
Such automation consists for instance in predicting the optimal planned dose with image-to-image deep convolutional neural networks like U-net.
In these long workflows, though, the challenge is to evaluate how uncertain the prediction is and how it propagates till the endpoints the physicians consider to make their treatment decision.
Most treatment workflows feed on medical images, whose quality is key, and lately reconstruction methods rely more and more on inverse problem solving to overcome the pitfalls of classical algebraic reconstruction.
[INMA] 2020-09-15 (14h) : Blind source separation of linear-quadratic near-separable mixtures
At Euler building (room A.002) + MS Teams
Speaker :
Christophe Kervazo (UMons)
Abstract :
In blind source separation (BSS), the data sets are assumed to stem from the mixing of elementary signals, called sources. Although the BSS methods assuming linear mixing models have known a wide range of successes, we will depart in this presentation from this setting and focus on the more advanced linear-quadratic (LQ) mixing models, which also include additional products of order two of the sources. Precisely, we here extend the near-separable nonnegative matrix factorization (NMF) framework to LQ mixings, enabling to design an efficient two-step unmixing algorithm coined SNPALQ + PP. The algorithm is shown to be robust to noise under discussed conditions. SNPALQ + PP furthermore exhibits good separation qualities in realistic hyperspectral unmixing numerical experiments.
[INMA] 2020-06-09 (14:00) : Online prediction of COVID19 dynamics. Belgian case study
At MS Teams
Speaker :
Nesterov, Yurii
Abstract :
In this talk, we present a new axiomatic model of epidemic development, which is consistent with the very special features of COVID19. This is a discrete-time switching linear model for predicting the dynamics of total number of infected persons and concentration of the asymptomatic virus holders in population. A small number of its parameters can be tuned using the available real-time dynamic data on virus propagation. Therefore, this model provides us also with online prediction of the future. As an application example, we present the conclusions of this model for epidemic of COVID19 in Belgium in 2020. During several months, our predictions were exact, typically, within the accuracy of 0.5%.
[INMA] 2020-05-26 (14:00) : A 20-minute talk given by a student (2/2)
At MS Teams
Speaker :
S
Abstract : Derivative free optimization (DFO) is the study of « zero order methods » that use no derivatives to find (or approach) the optimum of a problem. Nowadays there are many optimization problems for which the derivative computation is very expensive or noisy or even impossible. Without the use of the derivative we cannot expect the same performance as the classical first-order methods, but it is still possible to have methods that are guaranteed to converge to some stationary point. This is the case of the directional direct-search methods that poll multiples points in different well-chosen direction at each iteration and choose as next iterate the one that (best) decreases the objective. A basic example of this is the coordinate search method. The seminar will focus on this set of direct-search derivative free methods and their convergence results.
Seminars
Subscribe (ICS)
INMA Contact Info
Mathematical Engineering (INMA)
L4.05.01
Avenue Georges Lemaître, 4
+32 10 47 80 36
secretaire-inma@uclouvain.be
Mon – Fri 9:00A.M. – 5:00P.M.
JOIN US