— In the News —
Rachel Thomas, a co-founder of fast.ai, offers a thoughtful counter-perspective to the recent article in the Harvard Business Review that compared the trustworthiness of algorithms and human decision-makers.
At the recent Black Hat cybersecurity conference, vendors boasted about how they're using machine learning and AI to help make the world a better place. These days, that's an important selling point - but the data their products train on could be a problem.
— Sponsored Link —
Vettery specializes in tech roles and is completely free for job seekers. Interested? Submit your profile, and if accepted onto the platform, you can receive interview requests directly from top companies growing their data science teams.
— Tools and Techniques —
Julia 1.0, a major release, was announced last week. Besides new features, this release comes with a commitment to API stability. That means future releases shouldn't break your code. This post outlines what you can expect from Julia 1.0 and if you want to explore further, the Julia home page is a good place to start.
What begins as a thought-experiment turns into an awesome project that eventually has Alexa understanding sign language! Check out the short video first. This post includes a discussion of how it came together and a high-level walk-through of the implementation. It's a fun read and a link at the end goes to the code repo.
If you haven't been keeping up with the evolving Tensorflow ecosystem, this is a great post by Cassie Kozyrkov that summarizes the key points of a talk from Google Cloud Next titled, "What’s New with TensorFlow?"
— Resources —
Data science is often said to be built on three pillars: domain expertise, statistics, and programming. In this interview, Hadley Wickham, the Chief Scientist at RStudio and prolific R developer, chooses the best books to help aspiring data scientists build computer science fundamentals.
— Data Viz —
Brilliant. This step-by-step climate example shows how to tweak a visualization to make the data say whatever you want it to say.
Data visualization experts have been arguing against the use of rainbow color scales for years. This new article in Scientific American explores the issues and introduces a new color scale that avoids some of the typical problems and is also better for people who are color blind. Includes a very useful collection of linked references.