Bayesian intelligence in Warwick

Posted in pictures, Statistics, Travel, University life, Wines with tags , , , , , , , , , , , , on February 18, 2019 by xi'an

This is an announcement for an exciting CRiSM Day in Warwick on 20 March 2019: with speakers

10:00-11:00 Xiao-Li Meng (Harvard): “Artificial Bayesian Monte Carlo Integration: A Practical Resolution to the Bayesian (Normalizing Constant) Paradox”

11:00-12:00 Julien Stoehr (Dauphine): “Gibbs sampling and ABC”

14:00-15:00 Arthur Ulysse Jacot-Guillarmod (École Polytechnique Fedérale de Lausanne): “Neural Tangent Kernel: Convergence and Generalization of Deep Neural Networks”

15:00-16:00 Antonietta Mira (Università della Svizzera italiana e Università degli studi dell’Insubria): “Bayesian identifications of the data intrinsic dimensions”

[whose abstracts are on the workshop webpage] and free attendance. The title for the workshop mentions Bayesian Intelligence: this obviously includes human intelligence and not just AI!

Le Monde puzzle [#1085]

Posted in Books, Kids, R with tags , , , , , on February 18, 2019 by xi'an

A new Le Monde mathematical puzzle in the digit category:

Given 13 arbitrary relative integers chosen by Bo, Abigail can select any subset of them to be drifted by plus or minus one by Bo, repeatedly until Abigail reaches the largest possible number N of multiples of 5. What is the minimal possible value of N under the assumption that Bo tries to minimise it?

I got stuck on that one, as building a recursive functiion led me nowhere: the potential for infinite loop (add one, subtract one, add one, …) rather than memory issues forced me into a finite horizon for the R function, which then did not return anything substantial in a manageable time. Over the week and the swimming sessions, I thought of simplifying the steps, like (a) work modulo 5, (b) bias moves towards 1 or 4, away from 2 and 3, by keeping only one entry in 2 and 3, and all but one at 1 and 4, but could only produce five 0’s upon a sequence of attempts… With the intuition that only 3 entries should remain in the end, which was comforted by Le Monde solution the week after.

50/50 photography competition [another public image]

Posted in Statistics with tags , , , , , on February 17, 2019 by xi'an

research outreach wants to improve my public image [ltd]

Posted in Books, University life with tags , , , , , on February 17, 2019 by xi'an

I have received this most bizarre email (links are mine):

Dear Dr. Christian Robert,

Please excuse the direct nature of this contact, however I would like to speak with you regarding your work on the Accelerating MCMC algorithms study.

Research Outreach work in collaboration with research teams assisting with their Public Outreach activity, through means of a professionally produced series of publications and articles aimed at a broader audience. I understand Public Outreach is a very important issue within the research community – where it is often difficult to obtain large numbers of views and interactions.

We are currently working on our April 2019 edition. This publication will feature a wide variety of science disciplines, including approximately 28 research articles. We would like you to be one of those featured. My suggestion is that we might create a new article, possibly covering some basic details of the published paper I have seen online, or indeed, covering new ground or the wider scope of your work. This would be an entirely new article written and developed by Research Outreach, in close collaboration with you.

Research Outreach distributes all content, publications and your individual article, across major Social Media platforms, as well as our own website. Your work will be seen by a large and quantifiable global audience. We are a non-profit community company, all content we produce is free to share and can be downloaded by any reader. Our website is entirely transparent and answers any questions you may have.

I have briefly looked over the details of your work and believe Research Outreach to be an effective platform to communicate your outreach requirements. We publish under the Creative Commons Licence and you will own the copyright to your published material, and we can send you a separate PDF download, to be used on your web-page, at events, conferences, or as outreach material for students etc.

Please have a look at our website for examples of the publications we produce and further detail on our objectives and services. I can assure you the entire process requires very little work on your part, less than 1 hour, over a 10-week process. Some of the services we provide are paid for, this means there is a cost to you if you decide to take part. I would be delighted to explain this and our editorial process. I understand your time is limited.

Rather than writing any further detail, could we please find 5 minutes to discuss? During our call, I can answer any questions you may have and explain in detail the requirements and cost for taking part and of course the benefits of doing so.

This sounds rather predatory to me, since I have to pay this company to produce a paper I (co?)authored, and of dubious academic worth if I spend less than one hour on the project. When looking at the issues on-line, the contents of the articles seem quite light, with glossy images not always related with the topic of the journal article. If not predatory, then a sort of advertising agency for academics?! Weird times…

 

take a random integer

Posted in Books, Statistics with tags , , on February 16, 2019 by xi'an

A weird puzzle from FiveThirtyEight: what is the probability that the product of three random integers is a multiple of 100? Ehrrrr…, what is a random integer?! The solution provided by the Riddler is quite stunning

Reading the question charitably (since “random integer” has no specific meaning), there will be an answer if there is a limit for a uniform distribution of positive integers up to some number . But we can ignore that technicality, and make do with the idealization that since every second, fourth, fifth, and twenty-fifth integer are divisible by and , the chances of getting a random integer divisible by those numbers are , , , and .

as it acknowledges that the question is meaningless, then dismisses this as a “technicality” and still handles a Uniform random integer on {1,2,…,N} as N grows to infinity! Since all that matters is the remainder of the “random variable” modulo 100, this remainder will see its distribution vary as N moves to infinity, even though it indeed stabilises for $N$ large enough…

say cheese [jatp]

Posted in Statistics with tags , , , , , , , , , , on February 16, 2019 by xi'an

A picture taken at Jericho Cheese, Little Clarendon Street, Oxford, that I took this week when I visited this cheesemonger for the first time, after several years of passing by its tantalizing display of British cheeses! It happens to have become my most popular picture on Instagram, ranking above the fiery sunrise over the Calanques, and the alignment of brains at the Institute of Brain in Paris!

undecidable learnability

Posted in Books, Statistics, Travel, University life with tags , , , , , , on February 15, 2019 by xi'an

“There is an unknown probability distribution P over some finite subset of the interval [0,1]. We get to see m i.i.d. samples from P for m of our choice. We then need to find a finite subset of [0,1] whose P-measure is at least 2/3. The theorem says that the standard axioms of mathematics cannot be used to prove that we can solve this problem, nor can they be used to prove that we cannot solve this problem.”

In the first issue of the (controversial) nature machine intelligence journal, Ben-David et al. wrote a paper they present a s the machine learning equivalent to Gödel’s incompleteness theorem. The result is somewhat surprising from my layman perspective and it seems to only relate to a formal representation of statistical problems. Formal as in the Vapnik-Chervonenkis (PAC) theory. It sounds like, given a finite learning dataset, there are always features that cannot be learned if the size of the population grows to infinity, but this is hardly exciting…

The above quote actually makes me think of the Robbins-Wasserman counter-example for censored data and Bayesian tail prediction, but I am unsure the connection is anything more than sheer fantasy..!
~