Vectors hold the data, matrices transform it, gradients say which way to improve.
- VectorA house becomes a list of numbers: size, rooms, age.
- MatrixA layer multiplies the vector by a matrix of weights.
- LossOne number says how far the prediction is from the real price.
- GradientThe slope of the loss says which weights to nudge, and which way.
Explain 1/5
Source: Google ML Crash Course