Learning Machines: Foundations of Trainable Pattern-classifying SystemsMcGraw-Hill, 1965 - 137 sivua |
Kirjan sisältä
Tulokset 1 - 3 kokonaismäärästä 9
Sivu 83
... initial weight vector Ŵ1 . Since for each Ŷ , in Sŷ and Ŵ ; in Sŵ , Ŷ ; • Ŵ ; ≤ 0 , we have from Eq . ( 5.8 ) Ŵk + 1 = 1 Ŵ1 + Ŷ1 + Ŷ1⁄2 + + Ŷk 2 ( 5.9 ) We shall prove the theorem for the case W1 = 0 , although essentially the same ...
... initial weight vector Ŵ1 . Since for each Ŷ , in Sŷ and Ŵ ; in Sŵ , Ŷ ; • Ŵ ; ≤ 0 , we have from Eq . ( 5.8 ) Ŵk + 1 = 1 Ŵ1 + Ŷ1 + Ŷ1⁄2 + + Ŷk 2 ( 5.9 ) We shall prove the theorem for the case W1 = 0 , although essentially the same ...
Sivu 88
... initial weight vectors ; Y belongs to one of the train- ing subsets , say Yi . Then , either ( a ) W , ( k ) . Y > W1 ( k ) . Yk k j = 1 , . " R , ji or ( b ) there exists some l = 1 , R , li for which W ( ) . YkW , ( k ) . Yk j = 1 , R ...
... initial weight vectors ; Y belongs to one of the train- ing subsets , say Yi . Then , either ( a ) W , ( k ) . Y > W1 ( k ) . Yk k j = 1 , . " R , ji or ( b ) there exists some l = 1 , R , li for which W ( ) . YkW , ( k ) . Yk j = 1 , R ...
Sivu 91
... initial weight vector W1 we may remove from the training sequence those patterns Y ' for which W Yk ' > 0 . The reduced training sequence Sy then creates a reduced weight - vector sequence Sŵ such that k Ŷx • Ŵx ≤ 0 k ( 5.37 ) for all ...
... initial weight vector W1 we may remove from the training sequence those patterns Y ' for which W Yk ' > 0 . The reduced training sequence Sy then creates a reduced weight - vector sequence Sŵ such that k Ŷx • Ŵx ≤ 0 k ( 5.37 ) for all ...
Sisältö
Preface vii | 11 |
PARAMETRIC TRAINING METHODS | 43 |
SOME NONPARAMETRIC TRAINING METHODS | 65 |
Tekijänoikeudet | |
5 muita osia ei näytetty
Muita painoksia - Näytä kaikki
Yleiset termit ja lausekkeet
adjusted apply assume bank called cells changes Chapter classifier cluster column committee machine components consider consists contains correction corresponding covariance decision surfaces define denote density depends described discriminant functions discussed distance distributions elements equal error-correction estimates example exist expression FIGURE fixed given implemented important initial layered machine linear machine linearly separable lines majority matrix mean measurements modes negative networks nonparametric normal Note optimum origin parameters partition pattern hyperplane pattern space pattern vector pattern-classifying piecewise linear plane points positive presented probability problem properties PWL machine quadric regions respect response rule selection separable sequence side solution space Stanford step subsidiary discriminant Suppose terns theorem theory threshold training methods training patterns training procedure training sequence training subsets transformation values weight vectors Y₁ zero
Viitteet tähän teokseen
A Probabilistic Theory of Pattern Recognition Luc Devroye,László Györfi,Gabor Lugosi Rajoitettu esikatselu - 1997 |