Learn Neural Networks With Go, Not Math
summary
Live code builds a feedforward neural network Go implementation from scratch with neurons as weight slices plus bias, a layer type, sigmoid activation, and mean squared error loss. Training uses explicit gradients and plain gradient descent/backprop to update weights; a normalization service and Garden/Kubernetes rebuild containers on code changes. Demo shows a loss curve that tapers as epochs run, though predictions remain off due to an implementation bug.
AI-written summary · about
chapters (8)
- 0:00Intro
- 2:35Predictor
- 4:24Mean Square Error
- 7:07How does training work
- 9:17How training works
- 18:46Making a Network
- 19:04Prediction
- 20:06Gradient
related talks
33mLearn Neural Networks With Go - Not Math!
06mLearn Neural Networks With Go—Not Math! (Interview)
50mDeep Learning with Go
43m(Deep) Learn You a Neural Net for Great Good!
44mMachine Learning Exposed
51mMachine Learning Exposed Deep and Reinforcement Learning
18mMachine Learning and Go
26mDeep learning in Go - Present and future