Deep Learning is Just LEGO: & Other Hands on Machine Learning Activities
summary
Claims that core deep learning concepts become tangible by building physical models using actual LEGOs, clay or Play-Doh, and 3D prints; demonstrates LEGO convolutions that reveal filter depth across RGB channels, a 3D-printed gradient hill that shows overshoot and momentum, and a backpropagation coloring sheet that forces path-by-path tracing. These activities increase engagement and accessibility in classrooms and meetings but do not solve abstractions that depend on high-dimensional matrix algebra.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 1:05What is convolution
- 2:18A Fantasia
- 4:21Gradient Descent
- 7:02Back Propagation
- 8:01Coloring Sheet
- 9:30Designing Activities
- 12:28Questions
related talks
58mAn introduction to Machine Learning using LEGO
58mAn introduction to Machine Learning using LEGO
51mDeep Learning - What is it and What It Can Do For You
43mLessons Learned Using Machine Learning in Java
48mLessons learned using Machine Learning in Java
42mDeep Learning approaches meet 3D data
27mDemystifying Deep Neural Networks
54mIntroduction to Understanding Deep Learning