far south

Paris-Saclay campus debated in Nature

The newly created entity of the Paris-Saclay University is featuring in two editorials of Nature of 03 November for reaching a high ranking in one of the many league tables purportedly summarising the academic achievements of universities by a single number. This entity is made of the much older Université d’Orsay and of aggregated research institutes like Ecole Normale (formerly) de Cachan, Institut d’Optique, or Centrale Supélec, incidentally and uninterestingly located nearby my home. As an aggregate of high quality institutions, it is thus little surprise that it achieves a sufficient critical mass to reach a high ranking. Were the nearby Institut Polytechnique de Paris integrated as well, the ranking would have been even higher. (Why the two adjacent campuses did not merge defies rationality, but can be explained by politics and the long-standing opposition between Universités and Grandes Écoles in the French academic landscape.) I thus think the Nature editorial about the dangers to “the well-being of those on the academic front line” brought by the quest for high rankings is missing the point. By a fair margin. Indeed, it mixes the financial and institutional efforts made by [former president] Nicolas Sarkozy in creating a single campus with the funding of this mostly pre-existing campus [and in dire need of renovations, as exemplified by the new math department]. And seems to see the more competitive grant system in France connected with this creation when the [somewhat controversial] Agence Nationale de la Recherche in charge of the public-funded grants has been around since 2005. And I find that the unceasingly growing mille-feuille of aggregates, conglomerates, unions, initiatives, &tc. happening in the French academic landscape [like Paris Dauphine joining PSL a few years ago, whose status was confirmed today] are both blurring the picture and reducing the efficiency of the maneuvers by multiplying the administrative structures without creating a sense of belonging to a common institution. Plus ça change…

assistant/associate professor position in statistics/machine-learning at ENSAE

ENSAE (my Alma Mater) is opening a new position for next semester in statistics or/and machine-learning. At the Assistant Professor level, the position is for an initial three-year term, renewable for another three years, before the tenure evaluation. The school is located on the Université Paris-Saclay campus, only teaches at the Master and PhD levels, and the deadline for application is 31 March 2020. Details and contacts on the call page.

Julyan’s talk on priors in Bayesian neural networks [cancelled!]

Next Friday, 13 March at 1:30p.m., Julyan Arbel, researcher at Inria Grenoble will give a All about that Bayes talk at CMLA, ENS Paris-Saclay (building D’Alembert, room Condorcet, Cachan, RER stop Bagneux) on

Understanding Priors in Bayesian Neural Networks at the Unit Level

We investigate deep Bayesian neural networks with Gaussian weight priors and a class of ReLU-like nonlinearities. Bayesian neural networks with Gaussian priors are well known to induce an L², “weight decay”, regularization. Our results characterize a more intricate regularization effect at the level of the unit activations. Our main result establishes that the induced prior distribution on the units before and after activation becomes increasingly heavy-tailed with the depth of the layer. We show that first layer units are Gaussian, second layer units are sub-exponential, and units in deeper layers are characterized by sub-Weibull distributions. Our results provide new theoretical insight on deep Bayesian neural networks, which we corroborate with simulation experiments.

 

Roberto Casarin’s talk at CREST tomorrow

My former student and friend Roberto Casarin (University Ca’Foscari, Venice) will talk tomorrow at the CREST Financial Econometrics seminar on

“Bayesian Markov Switching Tensor Regression for Time-varying Networks”

Time: 10:30
Date: 14 March 2019
Place: Room 3001, ENSAE, Université Paris-Saclay

Abstract : We propose a new Bayesian Markov switching regression model for multi-dimensional arrays (tensors) of binary time series. We assume a zero-inflated logit dynamics with time-varying parameters and apply it to multi-layer temporal networks. The original contribution is threefold. First, in order to avoid over-fitting we propose a parsimonious parameterisation of the model, based on a low-rank decomposition of the tensor of regression coefficients. Second, the parameters of the tensor model are driven by a hidden Markov chain, thus allowing for structural changes. The regimes are identified through prior constraints on the mixing probability of the zero-inflated model. Finally, we model the jointly dynamics of the network and of a set of variables of interest. We follow a Bayesian approach to inference, exploiting the Pólya-Gamma data augmentation scheme for logit models in order to provide an efficient Gibbs sampler for posterior approximation. We show the effectiveness of the sampler on simulated datasets of medium-big sizes, finally we apply the methodology to a real dataset of financial networks.