Sequential Minimal Optimization Trainer
This is an implementation of Sequential Minimal Optimization plus
the 2 modifications suggested by S.S. Keerthi, et al.:
http://guppy.mpe.nus.edu.sg/~mpessk/
Methods:
train Train a new SVM.
|
Methods
|
|
_F
_examine_example
_in_I_0
_in_I_1
_in_I_2
_in_I_3
_in_I_4
_take_step
train
|
|
|
_F
|
_F ( self, i )
|
|
|
_examine_example
|
_examine_example ( self, i2 )
|
|
|
_in_I_0
|
_in_I_0 ( self, i )
|
|
|
_in_I_1
|
_in_I_1 ( self, i )
|
|
|
_in_I_2
|
_in_I_2 ( self, i )
|
|
|
_in_I_3
|
_in_I_3 ( self, i )
|
|
|
_in_I_4
|
_in_I_4 ( self, i )
|
|
|
_take_step
|
_take_step (
self,
i1,
i2,
)
|
|
|
train
|
train (
self,
training_set,
results,
kernel_fn,
C,
epsilon,
update_fn=None,
)
train(self, training_set, results, kernel_fn, C, epsilon,
update_fn=None) -> SVM
|
Exceptions
|
|
ValueError, "I could not find positive and negative training examples"
|
|
|