Archive for Stockholm
Monte Carlo Methods in Stockholm 2026
Posted in Statistics, Travel, University life with tags KTH, KTH Royal Institute of Technology, Markov chain Monte Carlo, MCMC, Monte Carlo Statistical Methods, Stockholm, Sweden, Swedish e-Science Research Centre, theoretical foundations, workshop on September 29, 2026 by xi'an
NobAIl prAIzes
Posted in Books, pictures, Travel, University life with tags AI, AlphaFold, Amazon, artificial intelligence, Boltzmann machines, Botlzman models, ChatGPT, Deep Mind, Friedrich Nietzsche, Google, Hopfield networks, Ising model, Markov random field, MCMC algorithms, neural network, Nobel Prize, Nobel Prize in Litterature, protein folding, South Korea, Stockholm, Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel, Sweden, twitter, vegetarian diet on October 11, 2024 by xi'an
I am quite surprised that the NobAIl committee did not select ChatGPT‘s Sam Altman for their literature prize—since he made a reality of monkeys typing at random on a typewriter, hence bound to a.s. produce past and future masterpieces—, now betting on Twitter’s Dorsey, Williams and Stone for their peace prize— for achieving worldwide harmony if within each single-minded community in only 140 characters—, and Amazon’s Jeff Bezos for their Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel—for produing the ultimate monopoly of a single worldwide convenience store—, given their picks for physics—with spin glass Ising models that have always sounded to me like the worst possible illustration for MCMC techniques— and chemistry—for a deep learning predictor of protein structures, built by large teams and numerous CPU hours, achieving high success rates if not perfection, but prediction is not explanation, reminding me of Nietzsche’s “physics, too, is only an interpretation and exegesis of the world (to suit us, if I may say so!) and not a world-explanation”—… Keeping their sharp focus on AI’s, corporate funded research, and male recipients… (As coïncidences come, I am currently reading a book of the literature 2024 Nobel recipient, Han Kang, The Vegetarian, that I find stunning and immensely original!)
Nobels away
Posted in University life with tags awards, Karolinska Institute, Nature, Nobel Prize, Science, Stockholm, Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel, Sweden, The Guardian on October 4, 2023 by xi'an
A self-administred poll in Nature about the respondents’ feeling about the prizes, with only 15% thinking it would be preferable to have none. As an absurd concept of singling out one scientist out the flock of top scientists, of comparing chalk and cheese, of idealising “great men” (since winners are predominantly men) with the ensuing hue and cry when they prove fallible, of focussing general public attention on a cartoonesque lone ranger version of scientific research, &tc. And to complete the absurdity, this meaningless statistic of a most common Nobel age!
ABC in Svalbard, April 12-13 2021
Posted in Mountains, pictures, Statistics, Travel, University life with tags 80th parallel North, ABC, ABC in Helsinki, ABC in Svalbard, ABCruise, Approximate Bayesian computation, Arctic, boat, cruise, Helsinki, Iceland, Longyearbyen, midnight sun, Norway, Spitzberg, Stockholm, Svalbard, workshop on November 4, 2019 by xi'an
conditional noise contrastive estimation
Posted in Books, pictures, University life with tags Charlie Geyer, conference, ICML 2018, intractable constant, logistic regression, machine learning, noise contrasting estimation, Stockholm, Sweden on August 13, 2019 by xi'an
At ICML last year, Ciwan Ceylan and Michael Gutmann presented a new version of noise constrative estimation to deal with intractable constants. While noise contrastive estimation relies upon a second independent sample to contrast with the observed sample, this approach uses instead a perturbed or noisy version of the original sample, for instance a Normal generation centred at the original datapoint. And eliminates the annoying constant by breaking the (original and noisy) samples into two groups. The probability to belong to one group or the other then does not depend on the constant, which is a very effective trick. And can be optimised with respect to the parameters of the model of interest. Recovering the score matching function of Hyvärinen (2005). While this is in line with earlier papers by Gutmann and Hyvärinen, this line of reasoning (starting with Charlie Geyer’s logistic regression) never ceases to amaze me!
