32 | The Learning Algorithm - Simple Truths of Data Science & Education

Today's episode features Brig Lamoreaux, a Sr. Advanced Analytics and AI Solution Architect at Microsoft, and has spent a large part of his career building research and software behind adaptive learning and modeling student knowledge

Episode 32 of the The Data Binge podcast, hosted by Derek Russell, titled "32 | The Learning Algorithm - Simple Truths of Data Science & Education" was published on January 19, 2020 and runs 56 minutes.

January 19, 2020 ·56m · The Data Binge

0:00 / 0:00

Summary

Today's episode features Brig Lamoreaux, a Sr. Advanced Analytics and AI Solution Architect at Microsoft, and Brig's been with the company for about 4 years. Brig has roughly 20 years of software development and engineering experience. Before Microsoft Brig worked for Apollo Education Group for 10 years, where he spent his time building research and software behind adaptive learning and modeling student knowledge. Some of the systems Brig has been involved in developing and productizing feature novel advances in education including cognitive science theories and data driven personalization. Brig has also spent a lot of time directing, programming, and leading 12-week learning boot-camps where students build hands-on skills using real world examples and projects. Brig finds a lot of value in learning and teaching, and you'll learn more from is lessons around how people learn throughout the discussion.

Episode Description

Themes of the discussion:

-How to measure depth of knowledge, and the feedback loop of learning:

—Teaching basic concepts

—Developing examples that enforce these concepts (which involve different teaching methods)

—Providing measurement tools for what was learned (tests and interviews: "tell me about Java")

-Machine learning in its most basic form, the data science pipeline, and a real world AI project he's using to teach his children how to solve problems with technology

-How to run a successful machine learning project:

—Can you initially draw some correlations from a variable, and it's potential outcome?

—Is there a human practitioner in the world that can make a probable estimate of the outcome you are trying to achieve? Ie, are there "hunches" that you can measure your hypothesis against?

—What specific data sets are these practitioners using? This is critical data

-People and teams, that are NOT successful with machine learning projects. A description of the WRONG approach, and mindset.

Thanks for listening!

How to connect with Brig:

LinkedIn: https://www.linkedin.com/in/briglamoreaux/

Github: https://github.com/briglx - over 203 contributions in the last year.

Resources and mentions:

https://machinelearningmastery.com/how-to-work-through-a-problem-like-a-data-scientist/

https://towardsdatascience.com/a-beginners-guide-to-the-data-science-pipeline-a4904b2d8ad3

Books:

https://www.amazon.com/Just-Mercy-Story-Justice-Redemption/dp/081298496X/ref=sr_1_1?keywords=just+mercy&qid=1579358686&sr=8-1

https://www.amazon.com/dp/1537196472/ref=cm_sw_em_r_mt_dp_U_DOXiEb9D3DSFK