Day 47: Reading the Old AI Literature is Underrated

🕵️‍♂️ State of AI Report 2018 - hot off the press @soundboy⁩ and I have set out to capture a snapshot of the exponential progress in AI with a focus on developments in the past 12 months. It's built for everyone, no expertise required. Check it out! https://t.co/PbyUBS29Mc

— Nathan Benaich (@NathanBenaich) June 29, 2018

Staggeringly beautiful…
"200 Years of U.S. Immigration As the Rings on a Tree" 🖌 @pmcruz & @wihbey https://t.co/vS3ClDgNen #dataviz #infovis pic.twitter.com/gMFJqSTNLj

— Mara Averick (@dataandme) July 5, 2018

🙌 post by @albertocairo (and a topic @thomasp85 has me thinking about a lot these days)
👉 "Transitions in visualization" https://t.co/4QDT8QPKMw #dataviz pic.twitter.com/0BOj1uaPGV

— Mara Averick (@dataandme) July 5, 2018

We have significantly updated our preprint on Variance Networks! Training only variances of the weights w_ij ~ N(0, sigma_ij^2) leads to state-of-the-art performance. Now more explanation, practical and theoretical results. https://t.co/5KobBhmHLA https://t.co/WTAnJoX4H6

— Arsenii Ashukha (@arsashuha) July 5, 2018

Given a satellite image, machine learning creates the view on the ground - Interesting work https://t.co/2Cypnw6o4w

— Nando de Freitas (@NandoDF) July 5, 2018

SLIM can model a scene conditioned only on human-written language descriptions. This allows it to capture the semantic meaning of spatial relationships in language. @tmramalho @TomasKocisky @arkitus @karlmoritz https://t.co/nAvFBKcb61 pic.twitter.com/sikxhXaLCg

— DeepMind (@DeepMindAI) July 5, 2018

"In a famous paper, Shao (1993) showed that leave-one-out cross validation does not lead to a consistent estimate of the model"https://t.co/ZHnWIYaOVW

— Gael Varoquaux (@GaelVaroquaux) July 5, 2018

In November 2017, I created an openly editable spreadsheet of tech ethics curricula. Now it has almost 200 courses listed, and I've blogged some context. Continue to spread the words, and let's make it an even better resource! https://t.co/XGX2xpg3S4

— Casey Fiesler (@cfiesler) July 5, 2018

The slides from my ⚡️talk at @rencontres_R about {purrr} are online !
🔗 : https://t.co/YUPYW2V6lA#RR2018 #RStats #RStatsFR pic.twitter.com/pOd22Oxr4E

— Colin Fay (@_ColinFay) July 5, 2018

Fastest way to train on CIFAR10 - now available on a PC near you! 🙂

This is a 🔋 included docker setup featuring recent work by @GuggerSylvain & @jeremyphoward (training with AdamW and the 1 cycle policy). Very excited by the development!https://t.co/C5QpkDFOwH pic.twitter.com/JA2NkeL212

— Radek (@radekosmulski) July 5, 2018

I wrote a tutorial on leveraging #DeepLearning to build a powerful image search engine quickly. It includes a notebook to walk you through and a codebase to play with. Also comes with a shoutout to @jeremyphoward and his great class on the topic! https://t.co/fRJHpMcq3h

— Emmanuel Ameisen (@EmmanuelAmeisen) July 5, 2018

It's official: @thomasp85 has taken over gganimate!

This is a complete rewrite for a better grammar of animated graphics, which he'll be keynoting about next week at #UseR2018

Thanks to everyone who has used & contributed to gganimate; I'm excited for its future! #rstats pic.twitter.com/w0sfxlyzhU

— David Robinson (@drob) July 5, 2018

More generally, it’s plausible that most powerful ideas are best expressed in executable form.

And so unless academic journals (or their replacements in the political economy of science) become executable, more and more of the best ideas will come from outside academia

— michael_nielsen (@michael_nielsen) July 5, 2018

Tutorial: The practical application of complicated statistical methods to fill up the scientific literature with https://t.co/1qTbO3ESXC

— Andrew Gelman (@StatModeling) July 5, 2018

A useful paradox for data scientists to keep in mind:

Most cities are small, but most people live in large cities

This relates to analyses of e.g. user engagement: most of your users probably don't do much, but most of your engagement is from frequent users

— David Robinson (@drob) July 5, 2018

I love the #rstats community.
Someone is like, "oh hey peeps, I saw a big need for this mundane but difficult task that I infrequently do, so I created a package that will literally scrape the last bits of peanut butter out of the jar for you. It's called pbplyr."
What a tribe.

— Frank Elavsky ᴰᵃᵗᵃ ᵂᶦᶻᵃʳᵈ (@Frankly_Data) July 3, 2018

If data is the new oil then security breaches are the new oil spill..? 🤔 https://t.co/6SepefHRQR

— Smerity (@Smerity) July 5, 2018

This 1795 essay by Laplace is fun to read: https://t.co/AsuW3ngw9h

— Ian Goodfellow (@goodfellow_ian) July 5, 2018

When I publish an academic paper - I pay the journal.
When I read an academic paper - I pay the journal.
When I review an academic paper for a journal - I don't get paid by the journal. https://t.co/NnI9lyD8TL

— Yaniv (((Erlich))) (@erlichya) July 3, 2018

I think of the reasons spreadsheet software is more approcahable than R/Python for a lot of folks is that you can directly see & manipulate the data structure. Learning how to conceptualize data structures is a specialized skill that takes time to learn.

— Dr. Rachael Tatman (@rctatman) July 5, 2018