— In the News —
Fantastic read about how artificial intelligence has defined Google over the years. This is the third of a four-part series by Steve Levy about "Search's quiet transformation." The entire series is highly recommended.
Every year Edge.org poses a different question to provoke and inspire debate. This year's question, "What do you think about machines that think?" has generated a fascinating mix of responses. Among the highlights so far:
Facebook recently open-sourced a handful of libraries to help developers build bigger, faster deep learning models than existing tools allow. Here's a good overview of what the tools do, why they're important, and how to get them.
— Tools and Techniques —
Two weeks ago, Mikhil Buduma's article, Deep Learning in a Nutshell was by far, the most popular article in Data Elixir. Here's his equally worthwhile followup.
Nice discussion on Cross Validated about the hidden assumptions of the k-means algorithm.
Here's a gentle introduction to using Python for data mining.
— Resources —
Nice cheatsheet for data wrangling with R's dplyr and tidyr packages.
260 Data Scientists from around the U.S. have reported salary information to glassdoor.com. Here are the details.
— Data Viz —
Great article by Christopher Olah about using visualization for understanding neural nets. Christopher worked with Google's Deep Learning Research Group and his visualization work is inspired by Brett Victor of WorryDream. Highly recommended.
It's not uncommon to think that "good" visualizations are somehow visually dramatic and colorful. Here's a wiser approach, including examples and rationale.