— In the News —
A year ago this week, Shivon Zilis published an article that organized every artificial intelligence and machine learning startup she could find. At the time, there were 2,529 of them. There's been a LOT of activity in this space since then and for this fresh update, Shivon redraws the landscape and explores where things are going. This is a must read for everyone interested in machine intelligence startups.
When one of the world’s biggest cyber weapon manufacturers was hacked into earlier this year, 400 GB of its data, internal emails, invoices, and source code made its way onto the Internet. To find the needles in that haystack, this article dives into an exploration of the metadata. This is a fantastic read with useful visualizations along the way.
Online fashion shopping is big business and a company called Lyst is one of the key players. The backbone of Lyst's business is a deep learning system that's used to analyze images and personalize recommendations for its customers. In this interview, Lyst's lead data scientist discusses why typical recommendation engines don't work for Lyst and the innovations that were developed instead.
Nice overview of why cracking Go is important and how researchers are currently approaching it. And if you haven't already seen it, this video from Fog Creek does a fantastic job of explaining the game and why Go AI is hard.
— Tools and Techniques —
This guide presents thorough descriptions and possible solutions to many of the kinds of problems that you will encounter when working with data. Highly recommended.
Nice exploration of a dataset from Shooting Tracker. This presents a complete workflow, starting with cleaning the data and continues with summary plots, forecasting, and visualization. This is very well done and includes a GitHub repo.
Clearly presented and well-organized tutorial series for cleaning up noisy images of text. A variety of approaches are covered here, making this a strong introduction to working with image data.
— Data Viz —
The New York Times is widely regarded as having one of the best data visualization teams in the world. Here's a short-list of their best work this year.
The team at Graphiq Engineering explored a plethora of existing color palettes and discovered that most aren't suitable for data visualization. This is a great overview of what they found and how they went about designing new palettes specifically for complex visualizations. Be sure to check out the resources at the end of the article too.