Perceptron learning Learn by adjusting weights to reduce error on training set The squared error for an example with input x and true output y is EEn-hw( 2 Perform optimization search by gradient descent: ∂E ErTX 0Err a ∂w aw; =Br×adg-9四=0W》 =-Er×g(im)×xj Simple weight update rule: W,←-W+a×Er×g(im)×xj E.g.,+ve error increase network output increase weights on +ve inputs,decrease on -ve inputs Chapter 20,Section5 11Perceptron learning Learn by adjusting weights to reduce error on training set The squared error for an example with input x and true output y is E = 1 2 Err 2 ≡ 1 2(y − hW(x))2 , Perform optimization search by gradient descent: ∂E ∂Wj = Err × ∂Err ∂Wj = Err × ∂∂Wj y − g(Σnj = 0Wjxj) = −Err × g0(in) × xj Simple weight update rule: Wj ← Wj + α × Err × g0(in) × xj E.g., +ve error ⇒ increase network output ⇒ increase weights on +ve inputs, decrease on -ve inputs Chapter 20, Section 5 11