😊

[😊 @tejscript]

No, Springboard’s Data Science Career Track won’t give you a Career in Data Science https://t.co/3rot7DTurD #DataScience #Bootcamp #JobGuarantee

[Zaid زيد @zaidalyafeai]

AttentioNN series
1. Attention maps
https://t.co/UR7fqUbBSk
2. Attention in NMT
https://t.co/ZDFPHAiagX
3. Attention in Image Captioning
https://t.co/0ppyqdQ4pM
4. Attention in Transofrmer Models
4.1 https://t.co/GyDEdih65M
4.2 https://t.co/deC2ffN5i1

[hardmaru @hardmaru]

The Deep Learning Revolution and Its Implications for Computer Architecture and Chip Design

Paper from @JeffDean provides his views on specialized hardware for ML, how ML can play a role in chip design and manufacturing, and future ML research directions

https://t.co/UDhxwCysML

[hardmaru @hardmaru]

Code for Weight Agnostic Neural Networks is now released

In addition to the original code to reproduce experiments, we also release a modular version ideal for future experimentation that runs nicely on a single laptop.

paper https://t.co/AjLYp8UQj2
code https://t.co/hvZa7YwCFK

[hardmaru @hardmaru]

This work was led by Adam Gaier, who did a 3-month internship in the Google Brain team in Tokyo. This idea came out after a few drinks in Roppongi. He has done some fantastic work in the Neuroevolution area in the past and is active in the GECCO community. https://t.co/YW5m9It5UD

[Google AI @GoogleAI]

Can neural network architectures alone, without learning any weight parameters, encode solutions for a given task? We search for “weight agnostic neural network” architectures that can perform various tasks even when using random weight values. Learn more→https://t.co/KgyymGFaTa

[hardmaru @hardmaru]

@GoogleAI Thread with some discussion: https://t.co/BfPIUjI7nm

[〈 Berger | Dillon 〉 @InertialObservr]

the coolest demonstration of frame-rate matching i've ever seen..

[Colin Carroll @colindcarroll]

I implemented multivariate regression in 9(!) different Python probabilistic programming frameworks and gave a short, opinionated blurb about each. Fun to see similarities and differences in API design in this space! https://t.co/bxpAc99S9D

[hardmaru @hardmaru]

Before burning €1Bn on trying to simulate the human brain, wouldn’t it make sense to first prove that we can build simulation models of simple worms and insects?

Even after we managed to map out entire connectome of C.elegans or Drosophila, we cannot simulate them convincingly.

[Lynn Cherny @arnicas]

Generative Deep Learning, the book I’ve been waiting for. https://t.co/XxJS926SSf (h/t @hardmaru)

[François Chollet @fchollet]

I've been reading a lot of old AI papers lately -- ranging from 1950 to 2010. It's interesting how contemporary a lot of these still feel to this day. The fundamental questions are still the same, and we still don't have any solid answers.

[Yann LeCun @ylecun]

ConvNet+satellite imagery+weak supervision = freely available super accurate worldwide street maps.
Brought to you by Facebook AI.
Interesting read. https://t.co/w21CA8ggTN

[Hadley Wickham @hadleywickham]

I've been working on a total rewrite of dtplyr, which lets you take advantage of (almost all of) data.table's speed while using dplyr syntax. The package is still in development, but I'd love for you to try it out and let me know how it goes: https://t.co/MXogoZ2lvu #rstats

[Kevin Markham @justmarkham]

WOW! pandas 0.25 (released last week) allows you to select a plotting backend other than Matplotlib... in one line of code 👏

Plotting libraries still have to build support for this, but ultimately it enables interactive plots directly from pandas! 🐼

Watch the 5-minute demo 👇

[Nando de Freitas @NandoDF]

AI protein-folding algorithms solve structures faster than ever https://t.co/u3WoVkRthz

[shoniko.eth @shoniko]

@arkosiorek @sabour_sara @yeewhye @geoffreyhinton Do you think Capsules are a close approximation of grid cells? As @Numenta sees them.https://t.co/8CD9662jtU

[Adam R. Kosiorek @arkosiorek]

We recently developed a new, unsupervised version of capsule networks (with @sabour_sara, @yeewhye, and @geoffreyhinton). I hope that my new blog will make it easier to understand some ideas that led to this work. Enjoy!
https://t.co/c1MuHRKYOD

[Geoffrey Hinton @geoffreyhinton]

This is a much better version of capsules.