(Deep) Learn You a Neural Net for Great Good!
summary
Live code builds a fully-connected neural net from scratch using numpy in a RISE slideshow, trains it first with explicit Python for-loops and then wraps it as a Scikit-learn classifier. The claim is concrete: implement forward/backprop, compute gradients with the chain rule, and update weights; the demo shows the net learns the classic XOR problem that logistic regression fails on. The implementation runs and converges in the demo.
AI-written summary · about
chapters (13)
- 0:00Neural networks intro
- 1:15What is a neural net
- 2:59Gradient descent basics
- 4:00XOR problem & activation
- 7:13For-loop neural net
- 11:53Forward & backward passes
- 14:41Chain rule & backprop
- 20:01Generalizing with layers
- 27:10Scikit-learn style net
- 33:49Power of composable layers
- 35:16Testing with tanh pitfalls
- 36:20Visualizing activations
- 40:30Closing & Q&A
related talks
51mMachine Learning Exposed Deep and Reinforcement Learning
44mMachine Learning Exposed
43mA Gentle Introduction to Neural Networks
26mMachine Learning in the Browser with Deep Neural Networks
33mLearn Neural Networks With Go - Not Math!
27mDemystifying Deep Neural Networks
01h 50mNeural Networks: Walkthrough
37mCode First, Math Later: Learning Neural Nets Through Implementation and Examples