Éditeur(s) : HAL CCSD Springer
Résumé : Third International Workshop, MLMI 2012, Held in Conjunction with MICCAI 2012
International audience
Universal Nearest Neighbours (unn) is a classifier recently proposed, which can also effectively estimates the posterior probability of each classification act. This algorithm, intrinsically binary, requires the use of a decomposition method to cope with multiclass problems, thus reducing their complexity in less complex binary subtasks. Then, a reconstruction rule provides the final classification. In this paper we show that the application of unn algorithm in conjunction with a reconstruction rule based on the posterior probabilities provides a classification scheme robust among different biomedical image datasets. To this aim, we compare unn performance with those achieved by Support Vector Machine with two different kernels and by a k Nearest Neighbours classifier, and applying two different reconstruction rules for each of the aforementioned classification paradigms. The results on one private and five public biomedical datasets show satisfactory performance.
Machine Learning in Medical Imaging
Nice, France

hal-00958909
https://hal.archives-ouvertes.fr/hal-00958909
https://hal.archives-ouvertes.fr/hal-00958909/document
https://hal.archives-ouvertes.fr/hal-00958909/file/Miccai.pdf
DOI : 10.1007/978-3-642-35428-1_15

Éditeur(s) : HAL CCSD Oxford University Press (OUP)
Résumé : International audience
The interpretation of coseismic surface deformation measurements through inversion techniques is of major importance to understand the mechanical behaviour of a seismic fault. Dense geodetic data sets in the vicinity of the ruptured fault provide unique constraints on detailed fault geometry and slip distribution at depth, making them complementary to seismological data. Bam earthquake (Mw 6.6, 2003 December 26) induced surface deformation has been precisely mapped by Envisat ASAR interferometry and by subpixel correlation techniques applied to Spot-5 and ASAR amplitude images. These oblique and horizontal estimations of deformation have been completed with one levelling profile along the main road crossing the rupture from west to east. We process these data (separately and jointly) in a two-step inversion technique, within the elastic half-space theory framework. Our objective is to determine the dislocation model at depth that satisfies simultaneously all the geodetic constraints. Also, we estimate the relative contribution of each geodetic data set to this inversion process.We first use a stochastic direct approach called neighbourhood algorithm in order to estimate the average characteristics of the rupture, and their relative uncertainty. Constraining in this way the geometry of the ruptured fault, we then linearize the inverse problem and compute the slip distribution on the fault using a standard weighted least-square technique, assuming the solution is smooth to some degree. At each step, we discuss the optimal models, their stability as well as the relative influence of each data set on the derived models parameters. Our preferred model reveals a shallow dislocation on a quasi-vertical fault, slightly dipping towards east. The slip vector has a strike-slip component as high as 2 m, while the dip-slip component seems negligible.However, the estimation of the resolution matrices emphasizes the fact that the details of deep fault slip distribution remain out of the scope of this ill-conditioned inverse problem. Yet, our preferred model suggests a main dislocation limited at depth between 1 and 6 km. By contrast, the aftershocks observed in the months following the earthquake are located just beneath the estimated main shock.
ISSN: 0956-540X

hal-00314173
https://hal.archives-ouvertes.fr/hal-00314173
https://hal.archives-ouvertes.fr/hal-00314173/document
https://hal.archives-ouvertes.fr/hal-00314173/file/GJI-2007-169-849-briole.pdf
DOI : 10.1111/j.1365-246X.2007.03358.x