— In the News —
Great talk by Pedro Domingos, author of the recently released book, "The Master Algorithm." This talk is easy to follow and offers a good overview of machine learning strategies. Ultimately, Domingos is looking for an algorithm that is capable of deriving all knowledge - past, present and future - from data.
Shivon Zilis is an investor at Bloomberg Beta with a specific interest in startups that are based on data and machine intelligence. This article provides a framework for thinking about market opportunities and strategies in this space. Really great insights here.
It may be awhile before data is regularly archived in DNA but the potential is amazing! Compared with electronic and magnetic memory, DNA can store data far more efficiently and for much longer time periods. This is a fascinating technology.
— Tools and Techniques —
Google's TensorFlow is getting a lot of attention lately but there are a lot of open-source projects to choose from. Here's a shortlist of the most important, including linked references and insights about how each excels.
Scikit Flow is a simplified interface for TensorFlow that mimics Scikit Learn. This makes it easy to move on from one-liner machine learning. You can start by using fit/predict and use the TensorFlow APIs as you get comfortable. This interface has become popular fast.
— Data Viz —
There has been quite a lot of discussion around the web lately about data visualization "rules." Along with an overview of the best of the discussion so far, this post by Chad Skelton offers his own worthwhile insights and examples.
Andy Kirk's monthly roundup of data viz projects and tutorials is a great source for ideas and inspiration from around the web. If you're a data viz fan or practitioner, this is a must read.