life and death along the RER B, minus approximations

Posted in Statistics, Travel with tags , , , , , , , , , , , , , , , on June 30, 2015 by xi'an

viemortrerbWhile cooking for a late Sunday lunch today, I was listening as usual to the French Public Radio (France Inter) and at some point head the short [10mn] Périphéries that gives every weekend an insight on the suburbs [on the “other side’ of the Parisian Périphérique boulevard]. The idea proposed by a geographer from Montpellier, Emmanuel Vigneron, was to point out the health inequalities between the wealthy 5th arrondissement of Paris and the not-so-far-away suburbs, by following the RER B train line from Luxembourg to La Plaine-Stade de France…

The disparities between the heart of Paris and some suburbs are numerous and massive, actually the more one gets away from the lifeline represented by the RER A and RER B train lines, so far from me the idea of negating this opposition, but the presentation made during those 10 minutes of Périphéries was quite approximative in statistical terms. For instance, the mortality rate in La Plaine is 30% higher than the mortality rate in Luxembourg and this was translated into the chances for a given individual from La Plaine to die in the coming year are 30% higher than if he [or she] lives in Luxembourg. Then a few minutes later the chances for a given individual from Luxembourg to die are 30% lower than he [or she] lives in La Plaine…. Reading from the above map, it appears that the reference is the mortality rate for the Greater Paris. (Those are 2010 figures.) This opposition that Vigneron attributes to a different access to health facilities, like the number of medical general practitioners per inhabitant, does not account for the huge socio-demographic differences between both places, for instance the much younger and maybe larger population in suburbs like La Plaine. And for other confounding factors: see, e.g., the equally large difference between the neighbouring stations of Luxembourg and Saint-Michel. There is no socio-demographic difference and the accessibility of health services is about the same. Or the similar opposition between the southern suburban stops of Bagneux and [my local] Bourg-la-Reine, with the same access to health services… Or yet again the massive decrease in the Yvette valley near Orsay. The analysis is thus statistically poor and somewhat ideologically biased in that I am unsure the data discussed during this radio show tells us much more than the sad fact that suburbs with less favoured populations show a higher mortality rate.

Kamiltonian Monte Carlo [no typo]

Posted in Books, Statistics, University life with tags , , , , , , , , , , on June 29, 2015 by xi'an

kamilHeiko Strathmann, Dino Sejdinovic, Samuel Livingstone, Zoltán Szabó, and Arthur Gretton arXived a paper last week about Kamiltonian MCMC, the K being related with RKHS. (RKHS as in another KAMH paper for adaptive Metropolis-Hastings by essentially the same authors, plus Maria Lomeli and Christophe Andrieu. And another paper by some of the authors on density estimation via infinite exponential family models.) The goal here is to bypass the computation of the derivatives in the moves of the Hamiltonian MCMC algorithm by using a kernel surrogate. While the genuine RKHS approach operates within an infinite exponential family model, two versions are proposed, KMC lite with an increasing sequence of RKHS subspaces, and KMC finite, with a finite dimensional space. In practice, this means using a leapfrog integrator with a different potential function, hence with a different dynamics.

The estimation of the infinite exponential family model is somewhat of an issue, as it is estimated from the past history of the Markov chain, simplified into a random subsample from this history [presumably without replacement, meaning the Markovian structure is lost on the subsample]. This is puzzling because there is dependence on the whole past, which cancels ergodicity guarantees… For instance, we gave an illustration in Introducing Monte Carlo Methods with R [Chapter 8] of the poor impact of approximating the target by non-parametric kernel estimates. I would thus lean towards the requirement of a secondary Markov chain to build this kernel estimate. The authors are obviously aware of this difficulty and advocate an attenuation scheme. There is also the issue of the cost of a kernel estimate, in O(n³) for a subsample of size n. If, instead, a fixed dimension m for the RKHS is selected, the cost is in O(tm²+m³), with the advantage of a feasible on-line update, making it an O(m³) cost in fine. But again the worry of using the whole past of the Markov chain to set its future path…

Among the experiments, a KMC for ABC that follows the recent proposal of Hamiltonian ABC by Meeds et al. The arguments  are interesting albeit sketchy: KMC-ABC does not require simulations at each leapfrog step, is it because the kernel approximation does not get updated at each step? Puzzling.

I also discussed the paper with Michael Betancourt (Warwick) and here his comments:

“I’m hesitant for the same reason I’ve been hesitant about algorithms like Bayesian quadrature and GP emulators in general. Outside of a few dimensions I’m not convinced that GP priors have enough regularization to really specify the interpolation between the available samples, so any algorithm that uses a single interpolation will be fundamentally limited (as I believe is born out in non-trivial scaling examples) and trying to marginalize over interpolations will be too awkward.

They’re really using kernel methods to model the target density which then gives the gradient analytically. RKHS/kernel methods/ Gaussian processes are all the same math — they’re putting prior measures over functions. My hesitancy is that these measures are at once more diffuse than people think (there are lots of functions satisfying a given smoothness criterion) and more rigid than people think (perturb any of the smoothness hyper-parameters and you get an entirely new space of functions).

When using these methods as an emulator you have to set the values of the hyper-parameters which locks in a very singular definition of smoothness and neglects all others. But even within this singular definition there are a huge number of possible functions. So when you only have a few points to constrain the emulation surface, how accurate can you expect the emulator to be between the points?

In most cases where the gradient is unavailable it’s either because (a) people are using decades-old Fortran black boxes that no one understands, in which case there are bigger problems than trying to improve statistical methods or (b) there’s a marginalization, in which case the gradients are given by integrals which can be approximated with more MCMC. Lots of options.”

art brut

Posted in pictures, Travel, University life with tags , , , on June 28, 2015 by xi'an

tantan

the girl who saved the king of Sweden [book review]

Posted in Books, Kids, Travel with tags , , , , , , , , , , , on June 27, 2015 by xi'an

When visiting a bookstore in Florence last month, during our short trip to Tuscany, I came upon this book with enough of a funny cover and enough of a funny title (possibly capitalising on the similarity with “the girl who played with fire”] to make me buy it. I am glad I gave in to this impulse as the book is simply hilarious! The style and narrative relate rather strongly to the series of similarly [mostly] hilarious picaresque tales written by Paasilina and not only because both authors are from Scandinavia. There is the same absurd feeling that the book characters should not have this sort of things happening to them and still the morbid fascination to watch catastrophe after catastrophe being piled upon them. While the story is deeply embedded within the recent history of South Africa and [not so much] of Sweden for the past 30 years, including major political figures, there is no true attempt at making the story in the least realistic, which is another characteristic of the best stories of Paasilina. Here, a young girl escapes the poverty of the slums of Soweto, to eventually make her way to Sweden along with a spare nuclear bomb and a fistful of diamonds. Which alas are not eternal… Her intelligence helps her to overcome most difficulties, but even her needs from time to time to face absurd situations as another victim. All is well that ends well for most characters in the story, some of whom one would prefer to vanish in a gruesome accident. Which seemed to happen until another thread in the story saved the idiot. The satire of South Africa and of Sweden is most enjoyable if somewhat easy! Now I have to read the previous volume in the series, The Hundred-Year-Old Man Who Climbed Out of the Window and Disappeared!

Introduction to Monte Carlo methods with R and Bayesian Essentials with R

Posted in Books, R, Statistics, University life with tags , , , , , , on June 26, 2015 by xi'an

sales1Here are the  download figures for my e-book with George as sent to me last week by my publisher Springer-Verlag.  With an interesting surge in the past year. Maybe simply due to new selling strategies of the published rather to a wider interest in the book. (My royalties have certainly not increased!) Anyway thanks to all readers. As an aside for wordpress wannabe bloggers, I realised it is now almost impossible to write tables with WordPress, another illustration of the move towards small-device-supported blogs. Along with a new annoying “simpler” (or more accurately dumber) interface and a default font far too small for my eyesight. So I advise alternatives to wordpress that are more sympathetic to maths contents (e.g., using MathJax) and comfortable editing.

salesBessAnd the same for the e-book with Jean-Michel, which only appeared in late 2013. And contains more chapters than Introduction to Monte Carlo methods with R. Incidentally, a reader recently pointed out to me the availability of a pirated version of The Bayesian Choice on a Saudi (religious) university website. And of a pirated version of Introducing Monte Carlo with R on a Saõ Paulo (Brazil) university website. This may be alas inevitable, given the diffusion by publishers of e-chapters that can be copied with no limitations…

another borderline conference

Posted in Kids, University life with tags , , , , , , on June 25, 2015 by xi'an

Following yesterday’s surprise at the unpleasant conference business run by WASET, I was once again confronted today with conference fees that sound like an unacceptable siphoning of research funds and public money. One of my PhD students got earlier personally invited to present a talk at EUSIPCO 2015, a European signal processing conference taking place in Nice next September and she accepted the invitation. Now, contrary to yesterday’s example, this EUSIPCO 2015 is a genuine conference sponsored by several European signal processing societies. From what I understand, speakers and poster presenters must submit papers that are reviewed and then published in the conference proceedings, part of the IEEE Xplore on-line digital library (impact factor of 0.04). As the conference is drawing near, my student is asked to register and is “reminded” of small prints in the conference rules, namely that “at least one author per paper must register by June 19, 2015 at the full rate”, student or not student, which means a 300€ difference in the fees and has absolutely no justification whatsoever since the papers are only processed electronically…

eupiscoI checked across a few of the past editions of EUSIPCO and the same rip-off rule applies to those as well. I see no rational explanation for this rule that sounds like highway robbery and leads to the de facto exclusion of students from conferences… In fine, my student withdrew her paper and participation at EUSIPCO.

Bayesian computation: a summary of the current state, and samples backwards and forwards

Posted in Books, Statistics, University life with tags , , , , , , , , on June 25, 2015 by xi'an

“The Statistics and Computing journal gratefully acknowledges the contributions for this special issue, celebrating 25 years of publication. In the past 25 years, the journal has published innovative, distinguished research by leading scholars and professionals. Papers have been read by thousands of researchers world-wide, demonstrating the global importance of this field. The Statistics and Computing journal looks forward to many more years of exciting research as the field continues to expand.” Mark Girolami, Editor in Chief for The Statistics and Computing journal

Our joint [Peter Green, Krzysztof Łatuszyński, Marcelo Pereyra, and myself] review [open access!] on the important features of Bayesian computation has already appeared in the special 25th anniversary issue of Statistics & Computing! Along with the following papers

which means very good company, indeed! And happy B’day to Statistics & Computing!

Follow

Get every new post delivered to your Inbox.

Join 875 other followers