Thursday, March 8, 2018 

Machine Learning Learning

And artificial Intelligence Intelligence

2 Days Left to Subscribe to Versioning at the Special Rate

This is the first of what I hope will be many more Versioning updates focused on particularly important subjects - this time, machine learning and AI. Hopefully this will start many of you on the road to creating a tool that will take over the world! (Disclaimer: if you accidentally invent Skynet as a result of reading these links, I’m not to blame.)

Think Tank

Conceptual pieces

First up, Wait But Why’s post on the coming artificial intelligence revolution is typically excellent, a good first step when thinking about AI and its potential impact on the world [waitbutwhy].

A guide to how deep learning works, aimed at “everyone” [medium.freecodecamp].

Three ways to think about technology (and therefore AI) [cyberselves].

And a glossary of AI terms [thenextweb]. (Yes, I kinda jammed machine learning and AI together with this update.)

Present Tensor

Starting points and tutorials

5 ways to get started with machine learning [sitepoint].

Exercises, lessons, lectures from experts, case-studies: Google’s crash course in machine learning is pretty hefty[].

A beginner’s guide to the top 10 machine learning algorithms [kdnuggets].

A video guide to writing your first machine learning code [youtube/googledevelopers].

TensorFlow [tensorflow] is the open source machine learning network that might be your best bet early on, so here’s a TensorFlow tutorial full of examples to get you started [github/aymericdamien].

Helpful cheat sheets for Python-based machine learning [startupsventurecapital].

A simplified MultiAgent Python Snake game via deep reinforcement learning [youtube/crazymuse].

A web dev’s guide to machine learning in JavaScript [robinwieruch].

How to start trading cryptocurrency with machine learning [wildml]. Just in case you wanted to jam two very trendy ideas together!

I’m a Model, You Know What I Mean?

Tools and resources

The complete collection of Facebook’s downloads and projects [research.fb].

And another collection of TensorFlow models [github/sarasra].

Google’s Cloud TPUs offer faster machine learning [].

When your research starts to hit its stride, Lucid [github/tensorflow] is a collection of tools and infrastructure for research in neural network interpretability.

AI Can’t Believe It


Ubisoft has an AI to help devs catch bugs before they’re committed [wired].

Examples of AI doing amazing things [medium/archieai].

Meet Michelangelo, Uber’s machine learning platform [eng.uber].

Baidu has AI that can analyze and reproduce voices when given just 3.7 seconds of audio [motherboard.vice].

Alibaba will use AI to optimize traffic and alleviate congestion in Kuala Lumpur [qz].

A Google AI will predict the risk of heart disease by looking into your eyes [theverge].

Job seekers are now tasked with convincing an AI they deserve the role [theguardian].

And there’s increasing concern about so-called “black box” algorithms, with humans unable to understand the inner workings of algorithms and tools [eurozine].

Finally, because of course: MariFlow [youtube/sethbling] is a self-driving Mario Kart example using a recurrent neural network.

There’s your intelligence about artificial intelligence - this will be updated over time. And more of these, on different subjects, will come through to paid subscribers soon! Let me know what you’d like to see, and sign up so you don’t miss out!

Curated by Adam