— In the News —
Hot in the news this week... a group of tech luminaries pledged more than a billion dollars to pursue advanced AI research and give the results to the public. Here's a great overview of what they're thinking.
In the very near future, most drugs will have both a chemical and digital component, as every pill will have a companion mobile app that collects patient-specific data. As millions of people use these apps, there's going to be an incredible new data stream to mine. Fascinating article!
What if AI learned more like children? Gary Marcus, a psychology professor at New York University, recently founded a company called Geometric Intelligence to explore that idea. He seems to be on to something.
— Tools and Techniques —
Probability distributions are fundamental to statistics, just like data structures are to computer science. With hundreds of probability distributions to choose from, this article offers a solid place to start.
If you've been using TensorFlow, you'll be interested in this. TensorDebugger is a visual debugger for deep learning. It extends TensorFlow with breakpoints and real-time visualization of the data flowing through the computational graph. This has just been out a week and is gaining fans quickly.
Nice exploration of database options and where each excels.
What do you do with a 120GB csv file?! This is a great Reddit discussion with lots of options and links to consider.
— Data Viz —
Histography is a MUST-PLAY-WITH exploration of Wikipedia. In this short interview, Matan Stauber describes how it came together and offers advice for integrating data and design.
In this podcast episode of This Old New Business, Cole Nussbaumer Knaflic talks about her new book, Storytelling with Data: A Data Visualization Guide for Business Professionals. This is a worthwhile interview and is downloadable so you can listen anywhere.
— Inspiration —
The Second Annual Data Science Bowl was recently announced and it looks like a good one! This year's competition is putting data science to work in cardiology. The task is to examine MRI images to create an algorithm that automatically measures cardiac health. Along with serving a good cause, the documentation and tutorials for this look awesome.
Here's a good vehicle for getting your feet wet with Spark. Devpost is hosting a challenge to use weather data and Analytics for Apache Spark to build smart weather apps. Prizes total $30,000 USD.