— In the News —
Karen Hao from the MIT Technology Review downloaded every AI paper abstract from arXiv through mid-November, 2018 and studied the trends. Among the findings... the era of deep learning may be coming to an end.
— Sponsored Link —
Data is everywhere, and is now more important than ever. Companies need developers with data science skills to work on projects ranging from social media marketing to building powerful machine learning applications, and those who have the skills stand out. If you're looking to do more with your data, then this exclusive offer from Manning is the perfect way to get started.
— Tools and Techniques —
The latest version of fast.ai's Practical Deep Learning for Coders course has just been released and it's 100% new material. This course is highly regarded and doesn't cost anything. Also, Google's Colab is now supporting fast.ai directly, which means you can now run all the lessons for the course and do your own R&D with a GPU-powered Jupyter Notebook - for free.
Counterfactuals are probablilistic answers to "what would have happened if" questions. They're powerful but tend to be hard to grasp. In this third post of a series on causal inference, Ferenc Huszár takes a look at what counterfactuals are and how they can be useful.
Lukas Biewald, the founder of Weights & Biases, describes three key reasons that machine learning projects can fail spectacularly and offers practical suggestions to deal with these issues.
Zev Ross' summary of the recent rstudio::conf identifies 15 useful new tools and ideas for R users. Each of the 15 includes a brief overview and links to the details.
In this short post, the team at GitHub ranks the most commonly used languages and libraries with the "machine-learning" tag. Covers usage, which projects are growing the fastest, which have the most contributors, etc.
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.
— Resources —
If you either use datasets that you didn't create or you make your own datasets available to others, this is a great reference of commonly used licenses to know about. Each of the the 13 entries in this list includes a short summary of the license terms and a link to the details.
— Data Viz —
The latest chapter in the D3 in Depth series explores D3’s approach to rendering geographic information. D3 in Depth goes beyond introductory tutorials and is a fantastic resource for learning how this popular library works.