Global AI and Data Science

Combating Algorithmic Bias: Internal Auditing Tools + Processes, Datasets, and Potential Trouble Areas

Companies like Facebook are implementing internal tools and specialized teams to audit their products’ algorithms.  Popular datasets used for years as standards have recently been discovered to have bias. This summary article points out some recent news where bias in ML has incredible impact.

New! Rapid Prototyping in Machine Learning Course on Coursera

IBM just announced the course Machine Learning Rapid Prototyping with IBM Watson Studio on Coursera. This course is ideal for data scientists who have too many projects on their plate and need to save time. Like having your own data science assistant, Watson Studio AutoAI Experiments can automate some of your workflow, leaving you to focus on using your expertise for higher-level data science tasks. Read more

Full Stack Deep Learning

Some recognizable names in the ML space have come together to put out a course on shipping deep learning products. They don't try and teach the theory and math, but they will help you with deploying. Read more

Tech TV: Open Education for AI

Host Kinga Parrot talks to Aria Chernick, Associate Professor at the Innovation and Entrepreneurship Initiative and the Social Science Research Institute at Duke University. Aria is a major proponent of applying open-source principles to education in order to provide high-quality education for all. Read more

IBM Series on AI Trust: Now Available On-Demand, Two More Live Workshops in September

Watch this five-part series featuring IBM experts and sign up for either of the two remaining live workshops later this month. Read more

Deploying an Internal Browsable Knowledge Repository with Jupyter Notebooks

This summary is a write-up of an older article detailing how Airbnb works to preserve  knowledge shared internally. By making data-access easy, Airbnb makes it easy for decision-making across its org to be data-driven. Read more

Announcing the Consortium for Python Data API Standards

There are dozens of libraries in the Python ecosystem which all want to help data scientists be more effective with their work. The result of high interest levels and many groups investing time and energy is a fragmented set of standards across these data libraries. The Consortium for Python Data API Standards hopes to address that. Read more

Running Notebook Pipelines in JupyterLab

Imagine using a visual editor to manage the data science pipelines you're building using JupyterLab. Now stop imagining and start using Elyra - an OSS extension that brings key GUI-based capabilities to the libraries data scientists already love. Read more

Taming the Tail: Adventures in Improving AI Economics

The economics of AI solutions are complicated by the fact that AI companies are trying to build inroads to markets that software has had a hard time reaching. There's no cut and paste solution but the authors share lessons from their experience. Read more

Hopfield Networks is All You Need

"Hopfield networks can store exponentially many patterns ... and has exponentially small retrieval errors." The paper's authors published a paper. They then also wrote an in-depth blog. They also introduced a new PyTorch layer. Read more