Search Results

1 result for “history”

Formulae (1)

\mathbf{w} := \mathbf{w} + \eta\,(y-\hat{y})\,\mathbf{x}

The original learning rule that nudges weights when a prediction is wrong.