Archive for summary statistics

information maximising neural networks summaries

Posted in pictures, Statistics with tags , , , , , , , , on February 6, 2019 by xi'an

After missing the blood moon eclipse last night, I had a meeting today at the Paris observatory (IAP), where we discussed an ABC proposal made by Tom Charnock, Guilhem Lavaux, and Benjamin Wandelt from this institute.

“We introduce a simulation-based machine learning technique that trains artificial neural networks to find non-linear functionals of data that maximise Fisher information : information maximising neural networks.” T. Charnock et al., 2018
The paper is centred on the determination of “optimal” summary statistics. With the goal of finding “transformation which maps the data to compressed summaries whilst conserving Fisher information [of the original data]”. Which sounds like looking for an efficient summary and hence impossible in non-exponential cases. As seen from the description in (2.1), the assumed distribution of the summary is Normal, with mean μ(θ) and covariance matrix C(θ) that are implicit transforms of the parameter θ. In that respect, the approach looks similar to the synthetic likelihood proposal of Wood (2010). From which an unusual form of Fisher information can be derived, as μ(θ)’C(θ)⁻¹μ(θ)… A neural net is trained to optimise this information criterion at a given (so-called fiducial) value of θ, in terms of a set of summaries of the same dimension as the data. Which means the information contained in the whole data (likelihood) is not necessarily recovered, linking with this comment from Edward Ionides (in a set of lectures at Wharton).
“Even summary statistics derived by careful scientific or statistical reasoning have been found surprisingly uninformative compared to the whole data likelihood in both scientific investigations (Shrestha et al., 2011) and simulation experiments (Fasiolo et al., 2016)” E. Ionides, slides, 2017
The maximal Fisher information obtained in this manner is then used in a subsequent ABC step as the natural metric for the distance between the observed and simulated data. (Begging the question as to why being maximal is necessarily optimal.) Another question is about the choice of the fiducial parameter, which choice should be tested by for instance iterating the algorithm a few steps. But having to run simulations for a single value of the parameter is certainly a great selling point!

approximate likelihood perspective on ABC

Posted in Books, Statistics, University life with tags , , , , , , , , , , , , , , on December 20, 2018 by xi'an

George Karabatsos and Fabrizio Leisen have recently published in Statistics Surveys a fairly complete survey on ABC methods [which earlier arXival I had missed]. Listing within an extensive bibliography of 20 pages some twenty-plus earlier reviews on ABC (with further ones in applied domains)!

“(…) any ABC method (algorithm) can be categorized as either (1) rejection-, (2) kernel-, and (3) coupled ABC; and (4) synthetic-, (5) empirical- and (6) bootstrap-likelihood methods; and can be combined with classical MC or VI algorithms [and] all 22 reviews of ABC methods have covered rejection and kernel ABC methods, but only three covered synthetic likelihood, one reviewed the empirical likelihood, and none have reviewed coupled ABC and bootstrap likelihood methods.”

The motivation for using approximate likelihood methods is provided by the examples of g-and-k distributions, although the likelihood can be efficiently derived by numerical means, as shown by Pierre Jacob‘s winference package, of mixed effect linear models, although a completion by the mixed effects themselves is available for Gibbs sampling as in Zeger and Karim (1991), and of the hidden Potts model, which we covered by pre-processing in our 2015 paper with Matt Moores, Chris Drovandi, Kerrie Mengersen. The paper produces a general representation of the approximate likelihood that covers the algorithms listed above as through the table below (where t(.) denotes the summary statistic):

The table looks a wee bit challenging simply because the review includes the synthetic likelihood approach of Wood (2010), which figured preeminently in the 2012 Read Paper discussion but opens the door to all kinds of approximations of the likelihood function, including variational Bayes and non-parametric versions. After a description of the above versions (including a rather ignored coupled version) and the special issue of ABC model choice,  the authors expand on the difficulties with running ABC, from multiple tuning issues, to the genuine curse of dimensionality in the parameter (with unnecessary remarks on low-dimension sufficient statistics since they are almost surely inexistent in most realistic settings), to the mis-specified case (on which we are currently working with David Frazier and Judith Rousseau). To conclude, an worthwhile update on ABC and on the side a funny typo from the reference list!

Li, W. and Fearnhead, P. (2018, in press). On the asymptotic efficiency
of approximate Bayesian computation estimators. Biometrika na na-na.

computational statistics and molecular simulation [18w5023]

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

 I truly missed the gist of the first talk of the Wednesday morning of our X fertilisation workshop by Jianfeng Lu partly due to notations, although the topic very much correlated to my interests like path sampling, with an augmented version of HMC using an auxiliary indicator. And mentions made of BAOAB. Next, Marcello Pereyra spoke about Bayesian image analysis, with the difficulty of setting a prior on an image. In case of astronomical images there are motivations for an L¹ penalisation sparse prior. Sampling is an issue. Moreau-Yoshida proximal optimisation is used instead, in connection with our MCMC survey published in Stats & Computing two years ago. Transferability was a new concept for me, as introduced by Kerrie Mengersen (QUT), to extrapolate an estimated model to another system without using the posterior as a prior. With a great interlude about the crown of thorns starfish killer robot! Rather a prior determination based on historical data, in connection with recent (2018) Technometrics and Bayesian Analysis papers towards rejecting non-plausible priors. Without reading the papers (!), and before discussing the matter with Kerrie, here or in Marseille, I wonder at which level of precision this can be conducted. The use of summary statistics for prior calibration gave the approach an ABC flavour.

The hand-on session was Jonathan Mattingly’s discussion of gerrymandering reflecting on his experience at court! Hard to beat for an engaging talk reaching between communities. As it happens I discussed the original paper last year. Of course it was much more exciting to listen to Jonathan explaining his vision of the problem! Too bad I “had” to leave before the end for a [most enjoyable] rock climbing afternoon… To be continued at the dinner table! (Plus we got the complete explanation of the term gerrymandering, including this salamander rendering of the first identified as gerrymandered district!)

asymptotic properties of ABC now appeared

Posted in Books, Statistics, University life with tags , , , , , , on September 1, 2018 by xi'an

Our paper with David Frazier, Gael Martin and Judith Rousseau has appeared in print in Biometrika, Volume 105, Issue 3, 1 September 2018, Pages 593–607, almost exactly two years after it was submitted. I am quite glad with the final version, though, and grateful for the editorial input, as the paper clearly characterises the connection between the tolerance level ε and the convergence rate of the summary statistic to its parameter identifying asymptotic mean. Asymptotic in the sample size, that is.

ABC’ptotics on-line

Posted in Statistics with tags , , , , , , , on June 14, 2018 by xi'an

Our paper on Asymptotic properties of ABC with David Frazier, Gael Martin, and Judith Rousseau, is now on-line on the Biometrika webpage. Coincidentally both papers by Wentao Li and Paul Fearnhead on ABC’ptotics are published in the June issue of the journal.

Approximate Bayesian computation allows for statistical analysis using models with intractable likelihoods. In this paper we consider the asymptotic behaviour of the posterior distribution obtained by this method. We give general results on the rate at which the posterior distribution concentrates on sets containing the true parameter, the limiting shape of the posterior distribution, and the asymptotic distribution of the posterior mean. These results hold under given rates for the tolerance used within the method, mild regularity conditions on the summary statistics, and a condition linked to identification of the true parameters. Implications for practitioners are discussed.

Bayesian synthetic likelihood

Posted in Statistics with tags , , , , , , , on December 13, 2017 by xi'an

Leah Price, Chris Drovandi, Anthony Lee and David Nott published earlier this year a paper in JCGS on Bayesian synthetic likelihood, using Simon Wood’s synthetic likelihood as a substitute to the exact likelihood within a Bayesian approach. While not investigating the theoretical properties of this approximate approach, the paper compares it with ABC on some examples. In particular with respect to the number n of Monte Carlo replications used to approximate the mean and variance of the Gaussian synthetic likelihood.

Since this approach is most naturally associated with an MCMC implementation, it requires new simulations of the summary statistics at each iteration, without a clear possibility to involve parallel runs, in contrast to ABC. However in the final example of the paper, the authors reach values of n of several thousands, making use of multiple cores relevant, if requiring synchronicity and checks at every MCMC iteration.

The authors mention that “ABC can be viewed as a pseudo-marginal method”, but this has a limited appeal since the pseudo-marginal is a Monte Carlo substitute for the ABC target, not the original target. Similarly, there exists an unbiased estimator of the Gaussian density due to Ghurye and Olkin (1969) that allows to perceive the estimated synthetic likelihood version as a pseudo-marginal, once again wrt a target that differs from the original one. And the bias reappears under mis-specification, that is when the summary statistics are not normally distributed. It seems difficult to assess this normality or absence thereof in realistic situations.

“However, when the distribution of the summary statistic is highly irregular, the output of BSL cannot be trusted, while ABC represents a robust alternative in such cases.”

To make synthetic likelihood and ABC algorithms compatible, the authors chose a Normal kernel for ABC. Still, the equivalence is imperfect in that the covariance matrix need be chosen in the ABC case and is estimated in the synthetic one. I am also lost to the argument that the synthetic version is more efficient than ABC, in general (page 8). As for the examples, the first one uses a toy Poisson posterior with a single sufficient summary statistic, which is not very representative of complex situations where summary statistics are extremes or discrete. As acknowledged by the authors this is a case when the Normality assumption applies. For an integer support hidden process like the Ricker model, normality vanishes and the outcomes of ABC and synthetic likelihood differ, which makes it difficult to compare the inferential properties of both versions (rather than the acceptance rates), while using a 13-dimension statistic for estimating a 3-dimension parameter is not recommended for ABC, as discussed by Li and Fearnhead (2017). The same issue appears in the realistic cell motility example, with 145 summaries versus two parameters. (In the philogenies studied by DIYABC, the number of summary statistics is about the same but we now advocate a projection to the parameter dimension by the medium of random forests.)

Given the similarity between both approaches, I wonder at a confluence between them, where synthetic likelihood could maybe be used to devise PCA on the summary statistics and facilitate their projection on a space with much smaller dimensions. Or estimating the mean and variance functions in the synthetic likelihood towards producing directly simulations of the summary statistics.

at the Isaac Newton Institute [talks]

Posted in Statistics with tags , , , , , , , on July 7, 2017 by xi'an

Here are the slides I edited this week [from previous talks by Pierre and Epstein] for the INI Workshop on scalable inference, in connection with our recently completed and submitted paper on ABC with Wasserstein distances: