Academic Data Science
If you're an academic who does data science (solid definition provided below) or works with or employs data scientists, please read this excellent report by @staeiou et al: https://t.co/gO6kCMsAKZ pic.twitter.com/A2aV3ifJWu
— Mike Kearney📊 (@kearneymw) June 22, 2018
Meta Learning by the Baldwin Effect
Meta Learning by the Baldwin Effect: Exciting new work from @chrisantha_f and team! Meta-learning through evolution allows the use of non-differentiable fitness functions. They show that the Baldwin effect is superior to Lamarckian evolution in many cases. https://t.co/MzOMMjW1hj pic.twitter.com/gL28VSReG6
— hardmaru (@hardmaru) June 22, 2018
Writing Kanji
I trained a neural network to write Kanji. https://t.co/65EI8CoILS pic.twitter.com/CwBfjkNdtv
— hardmaru (@hardmaru) June 22, 2018
Tutorials
Excited to see @mpjme using @observablehq to teach! https://t.co/IQlz7WwlQl
— Mike Bostock (@mbostock) June 22, 2018
This phenomenon can now be reliably replicated as shown in the follow up paper from @lnsmith613. We're now working with @GuggerSylvain to provide guidelines for more models. https://t.co/d2cAE7dDP8 https://t.co/AcUEkQ3CPW
— Jeremy Howard (@jeremyphoward) June 22, 2018
Give your #shiny applications a hand-drawn look ! (thanks to Wired JS by @preetster 👌 ) #rstats #shinyAppreciation https://t.co/wAch0Iiwgw pic.twitter.com/Zv7iO8CKLA
— Victor Perrier (@_pvictorr) June 20, 2018
Occasionally you have to read data into #rstats where one observation is split across multiple rows. A handy pattern to fix is to mutate, fill, and filter pic.twitter.com/be7JQNscCE
— Hadley Wickham (@hadleywickham) June 22, 2018
#rstats Word Clouds
💬 fun code-through feat. rtweet and #tidytext!
— Mara Averick (@dataandme) June 22, 2018
"Awesome Twitter Word Clouds in R" 👩🎨 @LittleMissData https://t.co/uWUGu6HSdj #rstats #dataviz #SoDS18 pic.twitter.com/DGznFOzA1W
Altair
@jakevdp Here's my write up on interactive Altair charts: https://t.co/440rNn34yB#dataviz #python pic.twitter.com/EiVGa3Kr7J
— Matthew Kudija (@mkudija) June 22, 2018
Resources
Machine Learning Papers with Codehttps://t.co/Sfa9BXI4OO pic.twitter.com/76frkyB4ni
— ML Review (@ml_review) June 23, 2018
KeOps - KErnel OPerationS, on CPUs and GPUs, with autodiff and without memory overflows. Amazing lib to compute on the fly kernel-like operators with @PyTorch support, for large scale model fitting (eg with MMD or optimal transport loss functions) https://t.co/sHfDfRPc0u pic.twitter.com/pGXYTnkxFc
— Gabriel Peyré (@gabrielpeyre) June 22, 2018
We just launched a slick new preview and visualization for public data shared on Kaggle https://t.co/It6ycLshRw pic.twitter.com/CcgaZ3OHI0
— Ben Hamner (@benhamner) June 22, 2018
Track the Progress of NLP (NLP-progress)
Do you often find it cumbersome to track down the best datasets or the state-of-the-art for a particular task in NLP? I've created a resource (a GitHub repo) to make this easier. https://t.co/roV5pFzMQe
— Sebastian Ruder (@seb_ruder) June 22, 2018
Joblib 0.12
joblib 0.12 is out with a better process pool management that does not crash openmp, more efficient dask interop and support for fast LZ4 compression in joblib.dump/load: https://t.co/3un2C15tPN
— Olivier Grisel (@ogrisel) June 22, 2018
Miscellaneous
Who made this? 👏🏻...👏🏻...👏🏻 pic.twitter.com/TgTiYBpMMy
— Drew Conway (@drewconway) June 22, 2018
Everyone makes mistakes during data analysis. Literally everyone. The question is not what errors you make, it's what systems you put into place to prevent them from happening. Here are mine. [a thread because I'm sad to miss #SIPS2018]https://t.co/pOLfExrZoc
— Michael C. Frank (@mcxfrank) June 22, 2018
Adobe is using machine learning to make it easier to spot Photoshopped images https://t.co/npnp5FLel0
— Nando de Freitas (@NandoDF) June 22, 2018
Sick of the internet shouting factory? Looking for a more civil place to discuss the big issues? After five years of development, we welcome you to Kialo, a system designed for thoughtful debate. https://t.co/pMlzo4pXNA
— Kialo (@KialoHQ) November 22, 2017