## Archive for École Polytechnique

## Cauchy’s head [jatp]

Posted in pictures, University life with tags Académie des Sciences, Augustin Cauchy, École Polytechnique, Bourbon kings, Charles X, Collège de France, French history, Institut Henri Poincaré, jatp, legitimists, Louis-Philippe, statute on August 1, 2021 by xi'an## MCMC, variational inference, invertible flows… bridging the gap?

Posted in Books, Mountains, Running, Statistics, Travel, University life with tags acronym, alpine climbing, École Polytechnique, biking, Eric Moulines, flat tyre, Hamiltonian Monte Carlo, HMC, invertible flows, MCMC, Pic du Midi d'Ossau, prairie, Pyrennées, teaching, University of Warwick, variational autoencoders, variational inference, webinar on October 2, 2020 by xi'an**T**wo weeks ago, my friend [see here when climbing Pic du Midi d’Ossau in 2005!] and coauthor Éric Moulines gave a very interesting on-line talk entitled *MCMC, Variational Inference, Invertible Flows… Bridging the gap?,* which was merging MCMC, variational autoencoders, and variational inference. I paid close attention as I plan to teach an advanced course on acronyms next semester in Warwick. (By acronyms, I mean ABC+GAN+VAE!)

The notion in this work is that variational autoencoders are based on over-simple mean-field variational distributions, that usually produce a poor approximation of the target distribution. Éric and his coauthors propose to introduce a Metropolis step in the VAE. This leads to a more general notion of Markov transitions and a global balance condition. Hamiltonian Monte Carlo can be used as well and it improves the latent distribution approximation, namely the encoder, which is surprising to me. The steps of the Markov kernel produce a manageable transform of the initial mean field approximation, a random version of the original VAE. Manageable provided not too many MCMC steps are implemented. (Now, the flow of slides was much too fast for me to get a proper understanding of the implementation of the method, of the degree of its calibration, and of the computing cost. I need to read the associated papers.)

Once the talk was over, I went back to changing tires and tubes, as two bikes of mine had flat tires, the latest being a spectacular explosion (!) that seemingly went through the tire (although I believe the opposite happened, namely the tire got slashed and induced the tube to blow out very quickly). Blame the numerous bits of broken glass over bike paths.

## Monsieur le Président [reposted]

Posted in Books, Statistics, University life with tags Alexander Grothendieck, École Polytechnique, coronavirus epidemics, Emmanuel Macron, epidemiology, France, Jean-Michel Marin, severe testing, Société française de Statistique, survey sampling, tribune, Université de Montpellier on April 11, 2020 by xi'an**Let us carry out screening campaigns on representative samples of population!**

Mr President of the Republic, as you rightly indicated, we are at war and everything must be done to combat the spread of CODIV-19. You had the wisdom to surround yourself with a Scientific Council and an Analysis, Research and Expertise Committee, both competent, and, as you know, applied mathematicians, statisticians have a role to play in this battle. Yes, to predict the evolution of the epidemic, mathematical models are used at different scales. This allows us estimate the number of people infected in the coming weeks and months. We are at war and these predictions are essential to the development of the best control strategy. They inform political decisions. This is especially with the help of these items of information that the confinement of the French population has been decided and renewed.

Mr President we are at war and these predictions must be the most robust possible. The more precise they are, the better the decisions they will guide. Mathematical models include a number of unknown parameters whose values should be set based on expert advice or data. These include the transmission rate, incubation time, contagion time, and, of course, to initialize dynamic mathematical models, the number of covered individuals. To enjoy more reliable predictions, it is necessary to better estimate such crucial quantities. The proportion of healthy carriers appears to be a particularly critical parameter.

Mr President, we are at war and we must assess the proportions of healthy carriers by geographic areas. We do not currently have the means to implement massive screenings, but we can carry out surveys. This means, for a well-defined geographic area, to run biological tests on samples of individuals that are drawn at random and are representative of the total population of the area. Such data would come to supplement those already available and would considerably reduce the uncertainty in model predictions.

Mr. President, we are at war, let us give ourselves the means to fight effectively against this scourge. Thanks to a significant effort, the number of individuals that can be tested daily increases significantly, let’s devote some of these available tests to samples representative. For each individual drawn at random, we will perform a nasal swab, a blood test, let us collect clinical data and other items of information on its follow-up barriers. This would provide important information on the percentage of immunized French people. This data would open the possibility to feed mathematical models wisely, and hence to make informed decisions about the different strategies of deconfinement.

Mr. President, we are at war. This strategy, which could at first be deployed only in the most affected sectors, is, we believe, essential. It is doable: designing the survey and determining a representative sample is not an issue, going to the homes of the people in the sample, towards taking samples and having them fill out a questionnaire is also perfectly achievable if we give ourselves the means to do so. You only have to decide that a few of the available PCR tests and serological tests will be devoted to these statistical studies. In Paris and in the Grand Est, for instance, a mere few thousand tests on a representative population of individuals properly selected could better assess the situation and help in taking informed decisions.

Mr. President, a proposal to this effect has been presented to the Scientific Council and to the Analysis, Research and Expertise Committee that you have set up by a group of mathematicians at École Polytechnique with Professor Josselin Garnier at their head. You will realise by reading this tribune that the statistician that I am does support very strongly. I am in no way disputing the competence of the councils which support you but you have to act quickly and, I repeat, only dedicate a few thousand tests to statistics studies. Emergency is everywhere, assistance to the patients, to people in intensive care, must of course be the priority, but let us attempt to anticipate as well . We do not have the means to massively test the entire population, let us run polls.

Jean-Michel Marin

Professeur à l’Université de Montpellier

Président de la Société Française de Statistique

Directeur de l’Institut Montpelliérain Alexander Grothendieck

Vice-Doyen de la Faculté des Sciences de Montpellier

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

Posted in pictures, Statistics, Travel, University life with tags École Polytechnique, ENSAE, France, job opening, machine learning, Palaiseau, Paris-Saclay campus, position, Statistics, Université Paris-Saclay on March 10, 2020 by xi'an**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.