Archive for book reviews

Locus awards 2021

Posted in Statistics with tags , , , , on July 15, 2021 by xi'an

After the Hugos, the Locus(es)! Here are the winners for different categories

the rising [book review]

Posted in Statistics with tags , , , , , , , , , , on June 23, 2021 by xi'an

When I received this second volume of the Alchemy Wars, the rising, it was most fortunately a weekend, and I devoured it within the two days! As hinted at by the title, hence not truly a spoiler!, this book ends up with the rise of the robots, thanks to the main characters already there in the first volume, Jax (reXened Daniel) the freed robot, Bérénice [missing her acute accents] the French master spy (code name Talleyrand), and Longchamp the charismatic commander of the Montréal (renamed Marseille-in-the-West) fortress. While the author seems to have invested more in the language of the Dutch Empire than in the one of the remaining French exiled to Québec, I did not spot crimes de lèse majesté on my native language (except for the above accents). A mystery remains though as to how, when crossing the Atlantic ocean, fugitives end up in Honfleur, east Normandy, and far inside the Channel. Returning to the plot per se, while its pace is breathless, with the revolutions of the characters’ paths bringing them into predictable contacts, and the dialogues are still great, the recourse to a hidden subterranean complex irked me as usual, while the repeated escapes of Bérénice from certain death, capture, brainwash, are just too much, even with the help of dei ex machina. This second volume is also less into pondering the meaning of free will and freedom, even though the sad discovery by Jax (sorry, Daniel!) of Neverland being somewhat a mirror of Netherlands is well-thought. Now waiting for the last volume and another free wekend (or a trip to Marseille!).

mathematical theory of Bayesian statistics [book review]

Posted in Books, Statistics, Travel, University life with tags , , , , , , , , , , on May 6, 2021 by xi'an

I came by chance (and not by CHANCE) upon this 2018 CRC Press book by Sumio Watanabe and ordered it myself to gather which material it really covered. As the back-cover blurb was not particularly clear and the title sounded quite general. After reading it, I found out that this is a mathematical treatise on some aspects of Bayesian information criteria, in particular on the Widely Applicable Information Criterion (WAIC) that was introduced by the author in 2010. The result is a rather technical and highly focussed book with little motivation or intuition surrounding the mathematical results, which may make the reading arduous for readers. Some background on mathematical statistics and Bayesian inference is clearly preferable and the book cannot be used as a textbook for most audiences, as opposed to eg An Introduction to Bayesian Analysis by J.K. Ghosh et al. or even more to Principles of Uncertainty by J. Kadane. In connection with this remark the exercises found in the book are closer to the delivery of additional material than to textbook-style exercises.

“posterior distributions are often far from any normal distribution, showing that Bayesian estimation gives the more accurate inference than other estimation methods.”

The overall setting is one where both the sampling and the prior distributions are different from respective “true” distributions. Requiring a tool to assess the discrepancy when utilising a specific pair of such distributions. Especially when the posterior distribution cannot be approximated by a Normal distribution. (Lindley’s paradox makes an interesting incognito incursion on p.238.) The WAIC is supported for the determination of the “true” model, in opposition to AIC and DIC, incl. on a mixture example that reminded me of our eight versions of DIC paper. In the “Basic Bayesian Theory” chapter (§3), the “basic theorem of Bayesian statistics” (p.85) states that the various losses related with WAIC can be expressed as second-order Taylor expansions of some cumulant generating functions, with order o(n⁻¹), “even if the posterior distribution cannot be approximated by any normal distribution” (p.87). With the intuition that

“if a log density ratio function has a relatively finite variance then the generalization loss, the cross validation loss, the training loss and WAIC have the same asymptotic behaviors.”

Obviously, these “basic” aspects should come as a surprise to a fair percentage of Bayesians (in the sense of not being particularly basic). Myself included. Chapter 4 exposes why, for regular models, the posterior distribution accumulates in an ε neighbourhood of the optimal parameter at a speed O(n2/5). With the normalised partition function being of order n-d/2 in the neighbourhood and exponentially negligible outside. A consequence of this regular asymptotic theory is that all above losses are asymptotically equivalent to the negative log likelihood plus similar order n⁻¹ terms that can be ordered. Chapters 5 and 6 deal with “standard” [the likelihood ratio is a multi-index power of the parameter ω] and general posterior distributions that can be written as mixtures of standard distributions,  with expressions of the above losses in terms of new universal constants. Again, a rather remote concern of mine. The book also includes a chapter (§7) on MCMC, with a rather involved proof that a Metropolis algorithm satisfies detailed balance (p.210). The Gibbs sampling section contains an extensive example on a two-dimensional two-component unit-variance Normal mixture, with an unusual perspective on the posterior, which is considered as “singular” when the true means are close. (Label switching or the absence thereof is not mentioned.) In terms of approximating the normalising constant (or free energy), the only method discussed there is path sampling, with a cryptic remark about harmonic mean estimators (not identified as such). In a final knapsack chapter (§9),  Bayes factors (confusedly denoted as L(x)) are shown to be most powerful tests in a Bayesian sense when comparing hypotheses without prior weights on said hypotheses, while posterior probability ratios are the natural statistics for comparing models with prior weights on said models. (With Lindley’s paradox making another appearance, still incognito!) And a  notion of phase transition for hyperparameters is introduced, with the meaning of a radical change of behaviour at a critical value of said hyperparameter. For instance, for a simple normal- mixture outlier model, the critical value of the Beta hyperparameter is α=2. Which is a wee bit of a surprise when considering Rousseau and Mengersen (2011) since their bound for consistency was α=d/2.

In conclusion, this is quite an original perspective on Bayesian models, covering the somewhat unusual (and potentially controversial) issue of misspecified priors and centered on the use of information criteria. I find the book could have benefited from further editing as I noticed many typos and somewhat unusual sentences (at least unusual to me).

[Disclaimer about potential self-plagiarism: this post or an edited version should eventually appear in my Books Review section in CHANCE.]

récits du vieux royaume [book review]

Posted in Books, pictures with tags , , , , , , on April 18, 2021 by xi'an

Following my enthusiastic reading of Jaworski’s Gagner la Guerre at the start of the first lockdown last year (!), I read this March the short stories at the beginning of a single volume called Récits du Vieux Royaume, which I found even better for the care in the writing style, the originality and diversity of the stories, the strong connection with traditional folklore and with the woes and worries of rural people, some of which could have been those of my not-so-remote ancestors, the mostly subtle insertion of details on the . (The connection with role game scenarios is close to invisible here.) It is only when reaching the second half of this book that I realised it contained the book I had bought and read last year. Rather than additional jewels…

“Elle racontait des histoires anciennes, des chroniques séculaires, des légendes à demi oubliées, ensevelies dans un passé fabuleux. Elle racontait la Geste de Leodegar le Resplendissant, ses batailles, ses victoires, l’union des clans autour du jeune héros habité par le souffle d’un dieu. Elle racontait le Vieux Royaume à l’époque de sa splendeur, Chrysophée aux murailles dorées, la prospérité et l’harmonie des campagnes, les forteresses orgueilleuses des trois duchés. Aux heures froides de la nuit elle racontait parfois les heures terribles de la guerre des Grands Vassaux, les morts marchant mêlés aux vivants dans les armées de Malvern, Chrysophée incendiée dans le soir, les derniers héros de Leomance, de Kahad Burg et de Valanael, ivres d’horreur et de désespoir, livrant combat pour défendre la berge de la Listrelle..”

a journal of the plague year [are we there yet?!]

Posted in Books, Kids, Mountains, pictures, Travel with tags , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , on March 20, 2021 by xi'an

Read the next volume of the Witcher series, Baptism of Fire, with even less enthusiasm than for the previous one, as the momentum of the series seems to have stalled… (Despite reading some highly positive reviews.) Some dialogues are funny enough, along with progressive views not particularly common in fantasy, like the support of reproductive rights, incl. abortion (and even less supported in the home country of the author, Andrzej Sapkowski!). But overall, not much happening and too much infodump!

Baked Ethiopian lentils & spinach mix, to get along with a slow cooking Ethiopian beef stew. And cooked more Venetian dishes. And had a great Korean streetfood dinner at (or from) MamiBaba by Quinsou, near Montparnasse, with pajeon (the cousin to okonomiyaki!) and kimchee. Accompanied by a first attempt at baking a chocolate pie.

Watched a few episodes of Alice in Borderland, vaguely suggested as hearsay by my daughter, but despite the fascinating scenes of an empty Tokyo, the plot is not particularly engaging, the tricks towards solving the game often lame, and the characters are not developed at all. Then watched Kurosawa’s Creepy, a gripping if not particularly realist psychological thriller that was premiered at the 2016 Berlin Film Festival. And reminded me of the much more disturbing Losey’s The Servant

Read two further volumes of John Harvey’s Charlie Resnick, in a random order, volumes that I found in and returned to the exchange section in front of our library as usual. And which I found almost as good as the first one, with its insistence on the humanity of each of the characters rather than indulging in manicheism. References to jazz pieces got a wee bit annoying by the third volume… And there is a maximal number of rye bread sandwiches with Polish pastrami I can swallow!

Watched also for the first time the fascinating The Wild Goose Lake (南方车站的聚会 which translates as A Rendez-Vous at a Station in the South), by Diao Yinan, a 2019 Cannes Festival selection, a psychological and violent noir film taking place in Wuhan among local gangs, when a gang boss kills by mistake a policeman after a very gory episode. The classical story line of the chase à la A bout de souffle is both tenuous and gripping, with an painful attention to colour and lightings, most scenes taking place at night with ghastly lights, with an intentional confusion between gangs of criminals and groups of cops, the final scene in full daylight making everything else sounding like a bad dream. The two main characters are striking, with an outlandish swan-like actress Gwei Lun-Mei. This also led me to watch the earlier Black Coal Thin Ice, which I also found impressive in terms of filming [that makes the cold and snow in this Northern city almost perceptible!] and definition of characters, once again involving Gwei Lun-Mei as the central, almost mute, and doomed, woman, but puzzling in terms of psychology and scenarios. (The shootout in the gallery is plain ridiculous imho.)