— Insight —
This report from the Polis think-tank explores how AI is already being used in journalism and how newsroooms think about the implications. Given how fast the tech is moving, this is important work that spans 71 news organizations in 32 different countries.
Every organization has their own use-cases, pain points and perspectives regarding data. In these interview snippets, data pros at 11 tech companies share insights from their corner of the data world.
— Tools and Techniques —
These short summaries of Machine Learning and NLP research papers cover a wide variety of authors, topics and venues from the past couple of years. Includes key points, diagrams and links for each paper.
TileDB is a new database that's designed from the bottom-up for data science. This post explores key problems with current solutions and how TileDB's approach is a more effective way to store, update, analyze, and share large sets of diverse data.
Confident Learning is an emerging field for characterizing label noise, identifying errors and learning using datasets with noisy labels. In this post, Curtis G. Northcutt describes what it is exactly, practical applications and how it works. This post also introduces a new open-source Python package for cleaning labels called the cleanlab.
— Resources —
I was at the Ignite Conference a couple weeks ago and there is a LOT that's worthwhile at Microsoft these days. Seriously. Along with the new HoloLens 2 and Project Silica, here are a few key resources that are worth checking out:
- AI Business School - Learn about AI strategy, responsibility, and technology through these online learning paths that are tailored to a variety of industries.
- Azure Tips and Tricks - This collection of 230+ tips, videos and conference talks span the entire universe of the Azure platform.
- AI for Good - This site is the portal to Microsoft's AI for Good programs. Includes information about specific projects, apps, datasets, grants, research, etc. If you're interested in Tech for Good, this is a must-explore resource.
— Data Viz —
The scientific literature is riddled with bad charts and graphs, leading to misunderstanding and worse. This post offers practical guidelines for choosing charts and colors to help others (and you!) understand your research. Includes lots of visual examples and common mis-perceptions.
Beautiful presentation of the network of paper citations at Nature. This short video is nice introduction to the project. The interactive is here >>