Open Data Repositories
(related to health, medicine and epidemiology)
Created a list with open data sources in the domains of health, medicine and epidemiology. ICYMI: https://t.co/95MYMcLQvr
— Maarten van Smeden (@MaartenvSmeden) July 14, 2018
Suggestions and comments are welcome (see next tweet) pic.twitter.com/isXBHr5tWA
Research
TCAV
Did your neural net say "zebra" because of pixel (17, 153)? Or because of the stripes?
— Martin Wattenberg (@wattenberg) July 13, 2018
TCAV interprets neural nets with high-level concepts, not low-level features. See https://t.co/OiyDaXznkS and https://t.co/rjzoGrTkAv cc @_beenkim @jmgilmer @Carryveggies @bengiswex @viegasf pic.twitter.com/K5DXqeos5h
AtDelfi
There are many challenges for AI in games: playing games, generating game content, modeling players, assisting designers etc. We introduce a new challenge: generating tutorials for games. In other words, given a game, explain to a human how to play it.https://t.co/L3dPhRl8KP pic.twitter.com/zZl6BCgOd9
— Julian Togelius (@togelius) July 14, 2018
Learning Resources
PyData London 2018
Awesome #datascience videos on our #PyData keynote playlist, including:
— PyData (@PyData) July 14, 2018
"Making the Big Data ecosystem work together with Python" by Holden Karau — on the work being done to decrease the overhead of #bigdata in #Pythonhttps://t.co/qvIvKnDoJL
Subscribe to our channel for more!
Version Control w/ Git
ICYMI, another 🌟 git guide (by proj type):
— Mara Averick (@dataandme) July 15, 2018
“A Quick Intro to Version Control w/ Git & GitHub” by @jdblischak @emo_davenport & @gvwilsonhttps://t.co/CAPlGZjZm9 #git #github pic.twitter.com/SG26XoPcki
Multithreading and Multiprocessing in Python
My talk at #SciPy2018 on Multithreading and Multiprocessing in Python is now viewable on Youtube!https://t.co/uv9lVww8Fb
— David Liu (@triskadecaepyon) July 13, 2018
git grep
Today, at #scipy2018 sprints, I learned that not enough folks know about `git grep`
— Paul “π” Ivanov (@ivanov) July 14, 2018
Search the current checkout of only committed files, ignores other stuff. pic.twitter.com/1bZbZFQ5T7
cProfile
Tip #8: Want to speed up your code? First, identify the bottleneck. Profiler is your friend. In #Python, use cProfile (https://t.co/cSnFEnDiWx). pic.twitter.com/UPIV6zjWvH
— Jeong-Yoon Lee (@jeongyoonlee) July 14, 2018
CoordConv Pytorch Implementation
Pytorch implementation of CoordConv https://t.co/hRo7b2VdRg #deeplearning #machinelearning #ml #ai #neuralnetworks #pytorch
— PyTorch Best Practices (@PyTorchPractice) July 14, 2018
Bad Science
Buggy code is bad science. Poorly tuned benchmarks are bad science. Poorly factored code is bad science (hinders reproducibility, increases chances of a mistake). If your field is all about empirical validation, then your code *is* a large part of your scientific output.
— François Chollet (@fchollet) July 15, 2018