Day 29: Maven Moment for Microsoft?

One of the more fun deep learning demos I've ever watched.https://t.co/Hv98lOuTrg

thanks to @antgoldbloom for showing it to me.

Still waiting for a phone app that does keypoint detection and coaches me to be a skier, tennis player, climber, etc?

— Dan Becker (@dan_s_becker) June 15, 2018

texture synthesis infinite loop, based on The Great Wave off Kanagawa (Hokusai, ~1830) pic.twitter.com/aJUxJgqgNe

— Gene Kogan (@genekogan) June 17, 2018

Microsoft announces building ICE cloud. While I know what Microsoft is thinking (business), making ICE more efficient at facial recognition accelerates the cruel processes ripping apart families in my community. https://t.co/R3Sdbhmw6Q https://t.co/nEqiI8lXG9

— Lilly Irani (@gleemie) June 18, 2018

Wonder if this becomes a Maven moment for Microsoft. https://t.co/FYnJzP0Rbu

— Mark Bergen (@mhbergen) June 18, 2018

Boom! Marian v1.5.0 released.

Includes:
- Extensions from the WNMT shared task on efficiency https://t.co/myYIU9fWIK
- Optimized GPU-decoding for Transformer models.

See chart below for speed comparison to v1.4.0 (based on FAIR's post)@jekbradbury @alvations pic.twitter.com/Bl9LryvHLj

— Marian NMT (@marian_nmt) June 17, 2018

For anyone who asked for my slides on Efficient (and Interactive) Deep Learning with Humans in the Loop from Thursday at @TechnionLive (or who didn't but might like them). I posted them to my website here: https://t.co/5RAxQaTHIG

— Zachary Lipton (@zacharylipton) June 17, 2018

🐍 #Python family: if you ever want to inspect the source code for a module that you've imported, but don't want to waste time hunting around for the .py file:

import inspect
import pprint

pprint.pprint( inspect.getsource( <module_name>)) pic.twitter.com/8AMHNMDt0u

— 👩‍💻 DynamicWebPaige @ GOTO; Amsterdam 🌷🇳🇱 (@DynamicWebPaige) June 17, 2018

Right?! 😁 And for #rstats friends - if you want to obtain the source code for a function, all you have to do is:

<library_name>:::<function_name> pic.twitter.com/cxLqF3DzOZ

— 👩‍💻 DynamicWebPaige @ GOTO; Amsterdam 🌷🇳🇱 (@DynamicWebPaige) June 17, 2018

👨‍🏫 brilliance: Ss wrote tutorials by topics, and they're 🌟!
"UO Psychology Data Science Gallery" by @robchavezhttps://t.co/c7X5AiCP1b #rstats pic.twitter.com/eHekzqHTZP

— Mara Averick (@dataandme) June 18, 2018

This fig in the Appendix of @kjhealy's data-viz 📕 is great!
👉 "How to read an R help page"https://t.co/n2u8b0lhyP #rstats pic.twitter.com/RJHdAS0Uo1

— Mara Averick (@dataandme) June 17, 2018

My first project for @DataCamp launched today! https://t.co/mPBDCYeCUF
Thanks to @venturidb for all his help with the launch. If you like the #tidyverse or want to learn it, you'll love this project! It's meant to complement @drob's tidyverse course. #rstats #datascience

— Sam Tyner (@sctyner) June 13, 2018

new blog post: a casual case study on how to speed up #rstats code featuring xml2, rcpp and furrrhttps://t.co/KOACH1O8dy pic.twitter.com/63T0SMsxH7

— alex hayes (@alexpghayes) June 15, 2018

A #tidyverse simulation to demonstrate that if you wait for two heads in a row, it takes 6 flips on average, while you wait for a heads then a tails, it takes 4 flips on average

h/t @CutTheKnotMath #rstats pic.twitter.com/V0zgOmCy7t

— David Robinson (@drob) June 17, 2018

Also a base #rstats matrix approach pic.twitter.com/GV16iFGRoV

— David Robinson (@drob) June 17, 2018

I've been thinking about why this is counterintuitive, even (especially?) to people experienced in probability. I think it's because effects like linearity of expected value spoil us in being able to treat events as independent even if they aren't https://t.co/mJd8h4vVGy

— David Robinson (@drob) June 18, 2018

People often ask if academia can keep up in research with industry given the superior resources there. Honestly, 99% of *interesting* academic work *even in DL* requires < 16 GPUs. This is maybe 1/100th the cost of an electron microscope. Resources are a bit over-emphasized

— Zachary Lipton (@zacharylipton) June 17, 2018

Reflections of a Human Discriminator
“If a fake is so good that even after the most thorough and trustworthy examination its authenticity is still open to doubt, is it or is it not as satisfactory a work of art as if it were unequivocally genuine?” https://t.co/WdaKiXaTBo

— hardmaru (@hardmaru) June 17, 2018

They’re called engineers, jesus https://t.co/5pdDQvYo3u

— Florent Crivello (@Altimor) June 16, 2018

The biggest scientific question of our time: is a potato battery sufficient to power a deep neural network?? https://t.co/O6pldCwxco

— Eric Jang (@ericjang11) June 17, 2018

When I first began learning to code, I saw `public static void main` in Java and was super confused. I was also reading a book that spent a page explaining each keyword. JavaScript was a relief coming from that verbosity.

Well, today I wrote this line of code 🤷‍♀️ pic.twitter.com/6kzXcsYWRu

— Pranay Prakash (@pranaygp) June 15, 2018