In the News
GM buys self-driving tech startup for more than $1B
General Motors announced today they are buying Cruise, a 3 year-old startup with 40 people that develops an aftermarket kit to help buyers convert their cars into autonomous vehicles.
Deepmind Health
Deepmind unveiled two products to help improve the UK health system:
- Streams, a risk identification tool to help identify patients at risk of acute kidney injury
- Hark, a clinical task management app
A world where everyone has a robot: why 2040 could blow your mind
Nature article with interesting charts on the ever accelerating technological change and its impact.
Also in the news this week...
- AlphaGo leads 2-0 in its match agains world champion Lee Sedol. You're certainly already aware. Read more on AlphaGo and on AlphaGo's strengths and weaknesses
- SalesForce acquires PredictionIO to build up its Machine Learning cluster. More
- Self-driving Google car hits bus in first crash attributed to the self-driving car. More
- Google's AI gets its first art show. More
- Brain launches an AI engine that aims to replace search. Nothing less. More
Learning
PlaNet - Google's neural network with “superhuman” ability to determine the location of images
Google engineers trained a deep neural network on 91 million geolocated images from the Web to determine an image's location just by looking at its content. The network has become great not only at recognizing landmarks, but also at spotting small details that help infer where on earth a photo was taken. Interesting read.
Code and understand Deepmind's Neural Stack Machine
This tutorial teaches DeepMind's Neural Stack machine via a very simple toy example, a short python implementation.
Neural Turing Machines
Introduction to the theory behind Neural Turing Machines and the way these can be implemented with a nice framework on top of Lasagne.
Software tools & code
Caffe on Spark - open source
Deep Learning often needs large amounts of data. Yet making Deep Learning frameworks work on large clusters has been notoriously difficult. Yahoo solved the issue by developing CaffeOnSpark, its tool to let the Caffe framework run on Spark.
Some thoughts
Connecting the world with better maps
Facebook has analyzed billions of satellite images to build detailed maps of worldwide human settlements. The granularity of this data is much finer than what was previously available. Facebook promises to release the data soon.
About
This newsletter is a weekly collection of AI news and resources.
If you find it worthwhile, please forward to your friends and colleagues, or share on your favorite network!
Share on Twitter · Share on Linkedin · Share on Google+
Suggestions or comments are more than welcome, just reply to this email. Thanks!