Day 58: AutoML

Looks like my #SciPy2018 talk on AutoML is up on YouTube:https://t.co/5TaM6NDaCx

— Randy Olson (@randal_olson) July 16, 2018

Video of my keynote at the International Joint Conference on Artificial intelligence, delivered yesterday in Stockholm. https://t.co/xxiQs1CTld

— Yann LeCun (@ylecun) July 17, 2018

"An Opinionated Introduction to AutoML and Neural Architecture Search" by @math_rachelhttps://t.co/PyobFlVxxn pic.twitter.com/EzrcAnlnrq

— Jeremy Howard (@jeremyphoward) July 16, 2018

Slides from our #ACL2018 tutorial on Neural Semantic Parsing with Alane Suhr, @sriniiyer88, @nlpmattg, and @LukeZettlemoyer now online: https://t.co/YcMiXSlffW

— Pradeep Dasigi (@pdasigi) July 17, 2018

Slides for my talk available here:https://t.co/OclMGUZfp6
Discussed our work on explainable machine teaching and teaching forgetful learners. https://t.co/9Akl2nAnRD

— Yisong Yue (@yisongyue) July 16, 2018

Wrote basic logistic regression tutorial https://t.co/C0WMAHa3lO using my own nn library. feedback appreciated. #MachineLearning
Also, matplotlib rocks: pic.twitter.com/Ods6TAQf9E

— harshvardhan Gupta (@harvey_slash) July 16, 2018

I just found “Intro to AI” _very_ nicely drawn video course by Dan Klein and @pabbeel. This is _really_ eyes and brain candies! I cannot wait to dig in and learn how to be a better teacher!!! *feels absolutely excited* https://t.co/jq78usCdDp
A few drawings from the MDP section. pic.twitter.com/K3Xo68uq3X

— Alf (冷在) (@AlfredoCanziani) July 15, 2018

America is the world's largest weapons exporter. I was curious to see what this looks like over time, so I mapped the flows of arms transfers leaving the U.S. from 1950 to 2017. The data comes from @SIPRIorg 's Arms Transfers Database. Full video here: https://t.co/5h1I6B2Xlp pic.twitter.com/WlSW73ZbLT

— Will Geary (@wgeary) July 16, 2018

It's been 10 days since @realDonaldTrump's tariffs dropped. To celebrate, we at @MacroPoloChina released the #ChinaFootprint, an interactive data visualization of Chinese spending+investment on US soil. Enjoy! https://t.co/Gjv82sb1ZL pic.twitter.com/bYzzZEldit

— Matt Sheehan (@mattsheehan88) July 17, 2018

Ouch: “prestigious institutions had on average 65% higher grant application success rates and 50% larger award sizes, whereas less-prestigious institutions produced 65% more publications and had a 35% higher citation impact per dollar of funding” https://t.co/zRQ9oX0xpR

— Björn Brembs (@brembs) July 16, 2018

In the context of machine learning research, science and engineering are not distinct concepts. You don't do "science" by thinking very hard about platonic ML concepts and then publishing your thoughts. You do science by engineering systems that test small ideas, iteratively.

— François Chollet (@fchollet) July 15, 2018

My contribution to the “science” vs engineering debate. I think papers should be about explaining reproducible ideas clearly, not about winning datascience competitions. (In the ideal world ;) it should be okay to sacrifice performance for clarity, simplicity and reproducibility. https://t.co/xiBIuspCys

— hardmaru (@hardmaru) July 17, 2018

Not saying SOTA isn’t important, but for example a much simpler algorithm that gets say 85% on CIFAR-100, 60 PTB perplexity, or 900 for CarRacing-v0 is more useful than overly complicated methods that get 89%, 56 or 920, esp for adapting them for other completely different tasks.

— hardmaru (@hardmaru) July 17, 2018

Just read another one of a myriad of #DeepLearning papers and realized that the authors are reporting trivia, but covering it up with an obscene amount of deep learning fluff and burned GPU time... I shall tear it apart in my next blog post. Stay tuned.

— Filip Piekniewski (@filippie509) July 13, 2018

OK, as promised here is a small autopsy of a paper I recently came across: https://t.co/uv7Vm0W4YX #Deeplearning #AI https://t.co/6d0GafR84q

— Filip Piekniewski (@filippie509) July 15, 2018

So true, unfortunately. In NLP, people have been adding position features to embeddings for a long time. I don’t think anyone has managed to write a whole paper about a one-line feature to make it sound academic. https://t.co/cb9AsRZWMO

— Denny Britz (@dennybritz) July 16, 2018

I really enjoyed the paper presentation, but the TLDR basically is: If your task cares about absolute positions it makes sense to add a position feature to your input! Do we need a paper for this?

— Denny Britz (@dennybritz) July 16, 2018

seaborn 0.9 is out! New plots, better aesthetics, improved documentation, and other novelties here: https://t.co/6FqFvVUBPh pic.twitter.com/AWqYZFGh3k

— Michael Waskom (@michaelwaskom) July 16, 2018