Multi-class Classification
Multi-class Classification
-
last sample videos actually a multiclass classification for differentiate 10 digits learning of handwriting recognition
-
multiclass is computed by extending the principal of one-vs-all
-
instead of classify y as earlier with number, classify based on matrices
-
(n+1) units x m classes
-
(5+1) x 10 = 60
-
+1 for bias unit
-
this is how to represent neural networks and hypothesis representation
-
next how to take training set and how automatically generated parameters for neural networks