Reading papers can result in surprises! A silly lifeblog entry:
Spotted this word2vec tutorial https://arxiv.org/abs/1402.3722 in Twitter (via https://twitter.com/zacharylipton/status/934590173752410113 ) and because we covered word2vec in our class last week, I read it; feeling studious, I decided to also look at the original Mikolov paper ( https://arxiv.org/abs/1310.4546 ) and check the eq 4 Goldberg&Levy talk about.
Enter the surprise (pic):
Hyvärinen & Gutmann both are UoH staff! I've seen them at the dept!
Read the (Finnish translation of) album Mickey's Craziest Adventures by Trondheim and Keramidas. Part of a series of Mickey adventures written and drawn by French artists, published by Glénat.
Was fun! The creators pretend that the comic was done in the 60s and they found it at a garage sale, so they can play with metafiction jokes (some pages are "missing" or "damaged"). Writing recognizably Trondheim; art is great.
@Altruest This is very true! I have accounts only on two relatively small instances and that already feels a bit too much.
Also, with multiple instances you have brand new problems like "under which accounts X, Z, Y I should reply to this toot I saw in my timeline on account Y". But I guess it works in the end. :P
(..also hello, I found your account by clicking through the #sciencefiction tag)
@panoramix I joined when a blogger wrote a blogpost "look at this cool thing called Mastodon, it's like a federated Twitter". A couple of another influxes that I have noticed have been when articles or blogposts about Mastodon have trended on HackerNews.
No doubt some of the influx phases are connected to Twitter crises...
@panoramix I believe "people hear about Mastodon" just one of those things that happens in phases. Something about how information spreads in human social networks.
@arttu Would you like to tell us what kind of things you find fascinating about Bayesian statistics?
The Economist on social media - another important argument for why #mastodon and other open platforms are so important.
Bayesian Optimization with Gradients
"Bayesian optimization has been successful at global optimization of expensive-to-evaluate multimodal objective functions. In this paper [the authors] show how Bayesian optimization can exploit derivative information to decrease the number of objective function evaluations required for good performance."
....and this time around I hope I managed to set up the visibility settings correctly...
More statistics stuff from Gelman's blog:
Advice for science writers: http://andrewgelman.com/2017/10/28/advice-science-writers/
"push [the authors] with hard questions: not just “Can you elaborate on the importance of this result?” but also “How might this result be criticized?”, “What’s the shakiest thing you’re claiming?”, “Who are the people who won’t be convinced by this paper?”
Favorite definition of statistical significance: http://andrewgelman.com/2017/10/28/favorite-definition-statistical-significance/
...does not involve p-values.
...and here's the mathoverflow link I forgot to add to previous toot: https://mathoverflow.net/questions/43690/whats-a-mathematician-to-do/44213#44213
Inspirational mathematics screed of the day, by Bill Thurston: "In short, mathematics only exists in a living community of mathematicians that spreads understanding and breaths life into ideas both old and new."
i also found out this evening that Ursula K Le Guin has a blog of sorts. it is decidedly not handled with a CMS, which makes it all the more wonderful.
Somehow this article reminds me of what I've read about the space race.
I'd like to be happy for the intern and say something cool and positive.
...but in reality I'm mostly thinking about how "if the exceptional people have headline-making publications under their belt before they even start their B.Sc, is this the standard by which I'll be judged too?"
s/Mastodon accounts/Mastodon instances ... sadly, the protocol probably does not support *editing* messages afterwards?
Applied math MSc student (U of Helsinki). inverse problems, statistics, ML.
The social network of the future: No ads, no corporate surveillance, ethical design, and decentralization! Own your data with Mastodon!