Archive for Edinburgh

guess what..?! Yet another worskhop in the endless summer Bayesian series!

Posted in Mountains, pictures, Running, Statistics, Travel, University life with tags , , , , , , , , , on April 18, 2018 by xi'an

Dennis Prangle pointed to me the perfectly timed i-like workshop taking place in Newcastle, on the days priors to ABC in Edinburgh and ISBA (similarly in Edinburgh!). (Note that Warwick is also part of the i-like network. Actually, the first i-like workshop was my first trip abroad after the Accident!) I may sound negative about these workshops, but on the opposite am quite a fan of them, just regretting that the main event did not take advantage of them all to reduce the volume of talks there. As I suggested, it could have been feasible to label these satellites as part of the main conference towards making speakers at these officially speakers at ISBA 2018 in case talks were required for support…

The i-like workshop 2018 is the sixth edition of a yearly series of workshops dedicated to the topic of intractable likelihoods, hosted by Newcastle University. The workshop will take place from Wednesday 20 June 2018 – Friday 22 June 2018 in Room 2.98, Armstrong Building, Newcastle upon Tyne. Registration is free and mandatory!

I spent a few days in Newcastle at the RSS meeting of 2013, with my friends Jim Hobert and Elias Moreno. Enjoying very much the city, its surroundings, the great meadow north of the city in a glorious sunset (I still bemoan not catching on camera!). And it is just in the vicinity of Hadrian’s Wall, just on the other side of the Borders, very close to Edinburgh in fact.

ABC and ISBA in Edinburgh [early registration deadline]

Posted in Books, Mountains, pictures, Statistics, Travel, University life, Wines with tags , , , , , , , , , , , on April 6, 2018 by xi'an

Just a reminder that the deadline for early registration at ISBA 2018, and hence at the satellite workshop ABC in Edinburgh, since there is alas no separate access to the later, is drawing near! Given the already rather sharp rates of £380 for ISBA members and £250 at this time, and the incoming rise (to plummet at $1,1000 for registering on site!!), it would be advised to register prior to April 15, if you intend to attend…

(Disclaimer: I am not part of the organising committee of either event. And I have already written several posts here and there complaining on the absurd inflation of conference fees.)

 

another instance of a summer of Bayesian conferences

Posted in pictures, Statistics, Travel, University life with tags , , , , , , , , , on March 15, 2018 by xi'an

As it happens, the next MaxEnt conference will happens in London, on 2-6 July, at the Alan Turing Institute, which makes it another perfect continuation of the ISBA meeting in Edinburgh, or of the Computational Statistics summer school in Warwick the week after. But in competition with BAYsm in Warwick and MCqMC in Rennes. I once attended a MaxEnt meeting in Oxford. (Oxford, Mississippi!) Which was quite interesting in the audience it attracted and the focus of the discussions, some of which were exhilaratingly philosophical!

ABCDE for approximate Bayesian conditional density estimation

Posted in Books, pictures, Statistics, Travel, University life with tags , , , , , , , , , , , , , on February 26, 2018 by xi'an

Another arXived paper I surprisingly (?) missed, by George Papamakarios and Iain Murray, on an ABCDE (my acronym!) substitute to ABC for generative models. The paper was reviewed [with reviews made available!] and accepted by NIPS 2016. (Most obviously, I was not one of the reviewers!)

“Conventional ABC algorithms such as the above suffer from three drawbacks. First, they only represent the parameter posterior as a set of (possibly weighted or correlated) samples [for which] it is not obvious how to perform some other computations using samples, such as combining posteriors from two separate analyses. Second, the parameter samples do not come from the correct Bayesian posterior (…) Third, as the ε-tolerance is reduced, it can become impractical to simulate the model enough times to match the observed data even once [when] simulations are expensive to perform”

The above criticisms are a wee bit overly harsh as, well…, Monte Carlo approximations remain a solution worth considering for all Bayesian purposes!, while the approximation [replacing the data with a ball] in ABC is replaced with an approximation of the true posterior as a mixture. Both requiring repeated [and likely expensive] simulations. The alternative is in iteratively simulating from pseudo-predictives towards learning better pseudo-posteriors, then used as new proposals at the next iteration modulo an importance sampling correction.  The approximation to the posterior chosen therein is a mixture density network, namely a mixture distribution with parameters obtained as neural networks based on the simulated pseudo-observations. Which the authors claim [p.4] requires no tuning. (Still, there are several aspects to tune, from the number of components to the hyper-parameter λ [p.11, eqn (35)], to the structure of the neural network [20 tanh? 50 tanh?], to the number of iterations, to the amount of X checking. As usual in NIPS papers, it is difficult to assess how arbitrary the choices made in the experiments are. Unless one starts experimenting with the codes provided.) All in all, I find the paper nonetheless exciting enough (!) to now start a summer student project on it in Dauphine and hope to check the performances of ABCDE on different models, as well as comparing this ABC implementation with a synthetic likelihood version.

 As an addendum, let me point out the very pertinent analysis of this paper by Dennis Prangle, 18 months ago!

dress down for ISBA 2018!

Posted in Mountains, pictures, Statistics, Travel, University life with tags , , , , , , on February 24, 2018 by xi'an

In  preparation for the ISBA meeting next June in Edinburgh, I bought a kilt! (A cheap acrylic version of the real thing, sold by a company called Tartinista. As shown next, when getting ready for a night-out at a Kenilworth curry restaurant by bearable if sub-zero temperatures, the tartan of the kilt is Black Watch, for a Highland Regiment.)

Although I do not foresee a huge follow-up (!), I’d still like to suggest we all wear kilts at the conference, both as a celebration of Scotland and as a gender neutral support for an all-inclusive ISBA. [Needless to say (?), let me stress here this is a purely personal initiative, with no backup whatsoever from Visit Scotland, the ISBA Exec, from the Program Committee, from the Local Organising Committee, or from safeISBA!]

 

weakly informative reparameterisations

Posted in Books, pictures, R, Statistics, University life with tags , , , , , , , , , on February 14, 2018 by xi'an

Our paper, weakly informative reparameterisations of location-scale mixtures, with Kaniav Kamary and Kate Lee, got accepted by JCGS! Great news, which comes in perfect timing for Kaniav as she is currently applying for positions. The paper proposes a unidimensional mixture Bayesian modelling based on the first and second moment constraints, since these turn the remainder of the parameter space into a compact. While we had already developed an associated R package, Ultimixt, the current editorial policy of JCGS imposes the R code used to produce all results to be attached to the submission and it took us a few more weeks than it should have to produce a directly executable code, due to internal library incompatibilities. (For this entry, I was looking for a link to our special JCGS issue with my picture of Edinburgh but realised I did not have this picture.)

ABC in Edinburgh

Posted in Books, Mountains, pictures, Statistics, Travel, University life, Wines with tags , , , , , , , , , , on January 25, 2018 by xi'an

As mentioned earlier on the ‘Og, there will be a satellite workshop ABC in Edinburgh, prior to the main ISBA meeting, taking place on Sunday, 24 June, 2018. The workshop is the last item of a series of “ABC in…” workshops on approximate Bayesian computation (ABC) and likelihood-free inference, which started with ABC in Paris in 2009! In this iteration, contributed talks and contributed posters can be submitted prior to May 1. (And there is a [extra] registration fee of 50 euros. And the deadline for early registration at ISBA 2018 is March 31, at the rather sharp rate of £380 for ISBA members.)The workshop is aimed at specialists and novices interested in statistical inference with complex models where exact computation of the likelihood function is not possible. The meeting will bring together researchers and practitioners in approximate Bayesian computation, likelihood-free inference, and related methods to discuss recent work on theoretical underpinnings, computational advances, and applications.

The invited speakers are

I am looking forward the workshop, having already booked my accommodation in the good City of Edinburgh.

(Disclaimer: I am not part of the scientific committee this round.)