Teaching kids ML

During the course of making my topic list for the Lindahl Letter this was the one that I was most worried about writing. Over the course of the last few years I have done a ton of machine learning and artificial intelligence training, but none of it was geared toward people without a strong foundation in mathematics and some statistics. My search started out with a few minutes of surfing the internet for content related to the topic at hand, “teaching kids machine learning.” The very first thing that jumped out at me was from the venerable International Business Machines (IBM) corporation.[1] They even have an office here in Denver and one not too far away in Boulder. That detail aside, the folks over at IBM have a course directly titled, “Machine Learning for Kids.” That course is free and has 1-4 hours of activity duration and is targeted at people ages 6 to 19+ years old. My big plan for later today is to see what happens when a member of my household within that age band gets to experience the course. More on that later…

Even KDnuggets, the training site, has been working on how to teach machine learning to kids.[2] They shared some links to some content and a little bit of information about training a model. It looks like Udemy has a course for, “Machine Learning for Kids and Beginners,” that has a cost of $14.99 to $29.99 depending on what type of sale is going on at the time.[3] Based on how that first 1-4 hour activity from IBM goes later today I may or may not kick off that Udemy course. They are super basic courses that I could very easily program in Colab notebooks using the free Google Colaboratory site from Google Research.[4] That made me wonder how many courses for kids might be hiding on Github or maybe on some website linking back to Colab notebooks. I’ll try to provide some feedback on these next week.

Links and thoughts:

Check out Yannic Kilcher this week, “[ML News] GitHub Copilot - Copyright, GPL, Patents & more | Brickit LEGO app | Distill goes on break”

This week Linus and Luke were really excited about GPUs on the WAN show, “Cheap GPUs are FINALLY Coming Back!! - WAN Show July 9, 2021”

Don’t miss out on this panel discussion from Machine Learning Street talk this week, “#56 - Dr. Walid Saba, Gadi Singer, Prof. J. Mark Bishop (Panel discussion)”

I spent about 20 minutes watching and thinking about this video, “BERT Model for Toxic Comment Classification (with GitHub Copilot's help)”

Top 5 Tweets of the week:

Dieter Bohn @backlon

Vergecast! ⁦@thedextriarchy⁩ joins to talk about the big Google antitrust suit and more. I talk a bit more about my Switch OLED model hands on. And ⁦@reckless⁩ has emotions about Jeep drones.

theverge.comVergecast: Google Play Store fees lawsuit, Nintendo’s OLED Switch, and an “EV Day” eventThe Vergecast discusses the newest antitrust lawsuit against Google, the Nintendo Switch OLED model, and all of the mottos that Stellantis announced for its “EV Day.”

Andrew Carr @andrew_n_carr

Deep learning is like Disneyland. From the outside it looks like a magical kingdom of wonder, but once you get there you realize it's expensive, crowded, and you're often just waiting a long time for things to happen

Dr. Nels Lindahl @nelslindahl

The @verge: GitHub’s automatic coding tool rests on untested legal ground. theverge.com/2021/7/7/22561… via @googlenews

theverge.comCan AI learn from any public code online?GitHub’s new Copilot tool has learned from public code online, but there isn’t a clear answer as to whether it’s legal. A future court decision could impact how AI algorithms learn from the open source development community.

Dr. Nels Lindahl @nelslindahl

‘The Great Resignation’: June’s US jobs report hides unusual trend

theguardian.com‘The Great Resignation’: June’s US jobs report hides unusual trendJune’s numbers suggest economy is continuing to recover at steady pace – but another pattern shows people are quitting their jobs

Dries Depoorter @driesdepoorter

🔥NEW PROJECT🔥 'The Flemish Scrollers' is software automatically tagging distracted Belgian politician when they use their phone on the daily livestreams. This with the help of AI. Results are posted on Twitter @FlemishScroller and IG TheFlemishScrollers driesdepoorter.be/theflemishscro…

Footnotes:

[1] https://www.ibm.org/activities/machine-learning-for-kids

[2] https://www.kdnuggets.com/2019/05/ai-machine-learning-kids.html

[3] https://www.udemy.com/course/machine-learning-for-kids-and-beginners/

[4] https://colab.research.google.com/notebooks/intro.ipynb

What’s next for The Lindahl Letter?

• Week 26: Machine learning as a service

• Week 27: The future of machine learning

• Week 28: Machine learning certifications? 

• Week 29: Machine learning feature selection

• Week 30: Integrations and your ML layer

I’ll try to keep the what’s next list forward looking with at least five weeks of posts in planning or review. If you enjoyed reading this content, then please take a moment and share it with a friend.