Université de Sherbrooke

About: Université de Sherbrooke is a education organization based out in Sherbrooke, Quebec, Canada. It is known for research contribution in the topics: Population & Receptor. The organization has 14922 authors who have published 28783 publications receiving 792511 citations. The organization is also known as: Universite de Sherbrooke & Sherbrooke University.

RNADE: The real-valued neural autoregressive density-estimator

Abstract: We introduce RNADE, a new model for joint density estimation of real-valued vectors. Our model calculates the density of a datapoint as the product of one-dimensional conditionals modeled using mixture density networks with shared parameters. RNADE learns a distributed representation of the data, while having a tractable expression for the calculation of densities. A tractable likelihood allows direct comparison with other methods and training by standard gradient-based optimizers. We compare the performance of RNADE on several datasets of heterogeneous and perceptual data, finding it outperforms mixture models in all but one case.

Discordance of monozygotic twins for thyroid dysgenesis: implications for screening and for molecular pathophysiology.

Abstract: Since the advent of biochemical screening for congenital hypothyroidism, the majority of monozygotic twins reported with thyroid dysgenesis have been discordant, and most were missed on neonatal screening, presumably due to fetal blood mixing. We hypothesized that there may be bias leading to preferential reporting of discordant twins and/or of false negative screening results. Therefore, we performed a systematic search for twins in two congenital hypothyroidism screening centers, Quebec and Brussels, that use a primary TSH approach. In Quebec, 10 pairs of twins were identified, all discordant for congenital hypothyroidism due to thyroid dysgenesis (4 monozygotic and 4 dizygotic pairs) and dyshormonogenesis (2 dizygotic pairs). The 6 pairs identified in the Brussels database were also all discordant for congenital hypothyroidism due to thyroid dysgenesis (1 monozygotic and 3 dizygotic pairs) and dyshormonogenesis (2 dizygotic pairs). The median increase in TSH between screening and diagnosis was 7-fold in the monozygotic twins vs. 2-fold in matched singletons (P = 0.02), suggesting fetal blood mixing between the twins. In summary, discordance for thyroid dysgenesis is the rule in monozygotic twins, and fetal blood mixing may result in delayed or missed diagnoses. We therefore conclude that 1) a second sample for congenital hypothyroidism screening at 14 d of age should be considered for all same-sex twins; and 2) thyroid dysgenesis generally results from epigenetic phenomena, early somatic mutations, or postzygotic stochastic events.

Neural autoregressive distribution estimation

Abstract: We present Neural Autoregressive Distribution Estimation (NADE) models, which are neural network architectures applied to the problem of unsupervised distribution and density estimation. They leverage the probability product rule and a weight sharing scheme inspired from restricted Boltzmann machines, to yield an estimator that is both tractable and has good generalization performance. We discuss how they achieve competitive performance in modeling both binary and real-valued observations. We also present how deep NADE models can be trained to be agnostic to the ordering of input dimensions used by the autoregressive product rule decomposition. Finally, we also show how to exploit the topological structure of pixels in images using a deep convolutional architecture for NADE.