The Firehose of Tensorflow in Week 4

Week 4 of the Fall 2018 Lancaster AI and AI Saturdays program happened on an absolutely beautiful day, 72 degrees Fahrenheit (22 Celcius), not a cloud in the sky, and somehow we still had a packed group of folks taking part in Machine Learning!

For our introductory level class, we focused on the first 5 parts of the Google Crash Course for Machine Learning, which took us from defining input features and labels to implementing a Linear Regressor directly using the Tensorflow Estimator API!

We found even at the higher abstraction levels of the Estimator, going from Python studies on Datacamp to Tensorflow was quite a jump in terms of complexity. Most people felt pretty good going through it, but there’s no doubt that we will also have to cover some coder-friendly Computer Vision examples like on PyImageSearch before the end of the course.

Wes Roberts and the Red and Black belts looked at multi-level perceptrons and the mathematical constructs behind back-propagation. Talk about some awesome foundations! This team went through quite a bit of calculus before closing up shop and planning on finishing the Fast.ai course level 2 for next week.


People’s final projects are starting to come into focus, which is great. We have people looking at open source home surveillance systems, Article 13 challenges, the Opioid crisis here in the United States, and more.

Can’t wait to see what people come up with after Week 5!