Professional Documents
Culture Documents
Lec 11 Single Layer Perceptron
Lec 11 Single Layer Perceptron
(3.7 ~ 3.9)
Learning-Rate Annealing Schedules(I)
When learning rate is large, trajectory may follow zigzagging
path
When it is small,procedure may be slow
Simplest learning rate parameter
( n) 0
Stochastic approximation
timevaring
when c is large, danger of parameter blowup for small n.
c
( n) (c is constant)
n
Learning-Rate Annealing Schedules(II)
Search then converge
schedule
in early stage, learning rate
parameter is approximately
equal to 0
for a number of iteration n
large compared to search
time constant ,
learning rate parameter
approximates as c/n
0
( n)
1 (n / )
(0 , is constant)
Perceptron(I)
Goal
classifying applied Input x1 , x2 ,..., xm into one of two classes
Procedure
if output of hard limiter is +1, to class C1 if it is -1, to class C2
input of hard limiter : weighted sum of input
m
v wi xi b
i 1
w x b 0
i 1
i i
w T ( n) x( n)
Perceptron Convergence Theorem(III)
Weight adjustment
if x(n) is correctly classified
w (n 1) w (n) if w T x(n) 0 and x(n) belongs to class C1
w (n 1) w (n) if w T x(n) 0 and x(n) belongs to class C 2
otherwise
n 2 2 n
w(n 1) x(k ) n max x(k )
2 2 2
2 X ( k )X 1
w0 k 1