Archive for Sceaux

a journal of the conquest, war, famine, death[s], and chaos year

Posted in Books, Kids, Mountains, pictures, Running, Travel with tags , , , , , , , , , , , , , , , , , , , , , , , , , , on April 10, 2024 by xi'an

Read John Scalzi’s Head On, which is set in the same future America as Lock In, involving again the Halden syndrome patients forced to live by remotely operating robots (threeps) and introducing an extreme form of American football adapted to these patients, since they cannot be injured when their threep is. (Not as in the terrible 1975 dystopic Rollerball, which was supposedly taking place in… 2018!). The two main FBI detectives are the same as in Lock In, with great and funny dialogues but with mostly dialogues!, and a surprising disregard for team work and reporting to their hierarchy. My conclusion of the review of Lock In thus stands:

“the Halden detective conveniently happens to be the son of a very influential ex-basketball-player and hence to meet all the characters involved in the plot. This is pleasant but somewhat thin with a limited number of players considering the issues at stake and a rather artificial ending.”

Starting to cook a matcha rice pudding as an experiment, which proved successful in keeping both the matcha taste ad the rice pudding texture, and in lowering considerably the input of sugar [from which I must shy] in the recipe. (In all honesty, I actually used an organic substitute to matcha, grown and made in China!)

Found out while going to a repair shop for a brake replacement that my second bike (the one that I can leave locked in the street for a few hours!) was in such a bad state that I should not drive it. The wheels had indeed lost most of their material at the level of the brakes, due to alien, abrasive, material getting stuck inside the brake pads. My nearby repair shop was clearly uninterested in repairing a cheap, ten year old, bike and gave me a quote that was larger than my original purchase amount. I thus found a Décathlon store nearby PariSanté campus and brought back a new wheel attached to my backpack, which proved more manageable than dreaded!

Watched in the nearby cinema A Man (ある男) by Kei Ishikawa, based on a book with the same title by Keiichiro Hirano, that won the Yomiuri Prize for Literature.  (The main actress Sakura Ando also played a central role in the fantastic Cannes Palme d’Or winner Shoplifters.) I went thinking it would be a psychological thriller, but it proved me wrong, as the movie is much more about self identity, intimacy, and societal prejudices, than a detective story about usurped identity. The pace is deliberately slow and the director light, impressionist, touch gives depth and freedom to the characters, while keeping some of the mysteries behind the story open. I really enjoyed the film, which was the first time I had returned to a cinema since watching a Jim Harrison documentary in 2022. I also discovered thanks to the beginning and final scenes an infinitely deep René Magritte’s painting, La Reproduction Interdite, which I had never seen, and which was a perfect still for the film message.

course du souffle au Parc de Sceaux [42:06, 10k, 128m⁺, 43/663, 1/19M5M]

Posted in pictures, Running with tags , , , , , , , , , , on September 26, 2023 by xi'an

All About that Bayes restart

Posted in pictures, Statistics, University life with tags , , , , , , , , , , on September 21, 2023 by xi'an

For all Bayesians and sympathisers in the Paris area, All about that Bayes seminars are restarting this semester with a talk by Kaniav Kamari (Centrale Supélec) on 10 October, 16h00, on Campus Pierre & Marie Curie, SCAI:

Bayesian principal component analysis

The technique of principal component analysis (PCA) has recently been expressed as the maximum likelihood solution for a generative latent variable model. In this talk, I’ll first present probabilistic reformulation that is the basis for a Bayesian treatment of PCA. Then, my focus will be on showing that the effective dimensionality of the latent space (equivalent to the number of retained principal components) can be determined automatically as part of the Bayesian inference procedure.

capture & maybe recapture

Posted in Books, pictures, Running, Statistics, University life with tags , , , , , , , , , on September 5, 2023 by xi'an

I read population size estimation with capture-recapture in presence of individual misidentification and low recapture arXived by Rémy Fraysse and coauthors on my flight back from Saigon. The setup is one of a capture-recapture experience where potential misidentification (of a recapture individual labelled as new) may occur due to visual identification errors as, e.g., in whale studies. Trying to handle the issue, Yoshizaki et al. (2011) proposed adding one layer to the temporal Darroch model M[t] via a probability α of creating a “ghost” (by failing to recognise a formerly observed individual). When representing the experiment as a partly observed Markov process (as in Dupuis, 1995), this addition brings another completely latent process for misidentification. Completely in the sense that a misidentification is never observed (while a proper identification is). This means there is an issue with identifiability between failing to capture and misidentification, as a new individual may be captured for the first time or misidentified on that round.

Processing this model (i.e., producing a simulation algorithm of the posterior) can be done formally by a Gibbs completion (as in Dupuis, 1995) but this may prove a non-irreducible scheme (Schofer & Bonner, 2015), a problem solved by considering instead Metropolis-Hastings steps. The current paper is an extension of the above to the multiple states Arnason-Schwarz model with no theoretical convergence issue, besides running a large sized completion. It is mostly a simulation experiment with a comparison on different priors on the misspecification rate, some highly informative, others not, with a frequentist assessment of coverage. Given the identifiability issue mentioned above, this is not particularly helpful since it simply exhibits the fact that with the right priors the parameter values are unbiasedly estimated, while a low recapture plus high misidentification setting makes estimation more difficult.