IMPORTANT UPDATE: With three courses under our belts, we just completed the first half of the sequence. But it's not too late to jump in and we want to make it easy for you. The next two weeks will be "catch up weeks." That means you can get started now and we'll help fast track you through the materials from the first half of the course, so you can continue with us when we begin course 4 on 4/18. 

Feature Engineering and Bias Detection is the subject of the 3rd and most recent course we've studied. Here's a brief summary of the topics we covered.

Week 1 of the course explored data transformation and feature engineering. In the study group we discussed:

• feature engineering and transformations in the context of the AI workflow
• employing the tools that help address class and class imbalance issues
• the ethical considerations regarding bias in data
• the dimension reduction techniques for both EDA and transformations stages
• topic modeling techniques in natural language processing
• topic modeling and visualization to explore text data

Week 2 focused on bias detection, in addition to best practices with regards to pattern recognition and data mining. I especially enjoyed getting to work with and discuss the AI Fairness 360 project. These tools are a real asset to practitioners as ML and AI find their way into applications that impact human behavior and decision making. Here's what the study group covered:

• The IBM AI Fairness 360 libraries to detect bias in models
• Best practices for handling outliers in high dimension data
• Employing detection algorithms as a quality assurance tool and a modeling tool
• Unsupervised learning techniques using pipelines as part of the AI workflow
• Using basic clustering algorithms

Check out the program page on the TWIML web site for more about this free program. To join, first register for the TWIML Community. You'll automatically be invited to our Slack group. After joining that, join the #ai_enterprise_workflow channel and you're in!

P.S. Feel free to reach out with any questions.

------------------------------
Sam Charrington
------------------------------

If you missed these sessions, you can catch the recordings here: