Material Detail
How a Neural Network Learns: gradient descent and backpropagation, interactive
A small neural network trains in the browser while a steel ball steps down a relief milled to the network's real error, one gradient step at a time. Choose a data set or place your own points, then set the learning rate, the points per step, momentum or Adam, and the network's width and depth; light in a glass-and-brass network runs forward with the numbers and back with each weight's share of the blame. Chapters cover the loss surface, step...
Show MoreQuality
- User Rating
- Comments
- Learning Exercises
- Bookmark Collections
- Course ePortfolios
- Accessibility Info