def fit(self, x, y, history):
self._svm = IncrementalMultiSVM(
dtype=x.dtype,
n_features=x.shape[1],
n_classes=2,
l2_regularization=self.l2_regularization,
n_sgd_iters=0,
bfgs_kwargs={
'maxfun': 1000,
'iprint': 0,
'm': 32,
'factr': 100
},
)
self._svm.fit(x, (y + 1) / 2, history)
class IncrementalSVM_MultiHack(object):
def __init__(self, l2_regularization):
self.l2_regularization = l2_regularization
def fit(self, x, y, history):
self._svm = IncrementalMultiSVM(
dtype=x.dtype,
n_features=x.shape[1],
n_classes=2,
l2_regularization=self.l2_regularization,
n_sgd_iters=0,
bfgs_kwargs={
'maxfun': 1000,
'iprint': 0,
'm': 32,
'factr': 100},
)
self._svm.fit(x, (y + 1) / 2, history)
def predict(self, x, history):
return self._svm.predict(x, history) * 2 - 1
class IncrementalSVM_MultiHack(object):
def __init__(self, l2_regularization):
self.l2_regularization = l2_regularization
def fit(self, x, y, history):
self._svm = IncrementalMultiSVM(
dtype=x.dtype,
n_features=x.shape[1],
n_classes=2,
l2_regularization=self.l2_regularization,
n_sgd_iters=0,
bfgs_kwargs={
'maxfun': 1000,
'iprint': 0,
'm': 32,
'factr': 100
},
)
self._svm.fit(x, (y + 1) / 2, history)
def predict(self, x, history):
return self._svm.predict(x, history) * 2 - 1
def fit(self, x, y, history):
self._svm = IncrementalMultiSVM(
dtype=x.dtype,
n_features=x.shape[1],
n_classes=2,
l2_regularization=self.l2_regularization,
n_sgd_iters=0,
bfgs_kwargs={
'maxfun': 1000,
'iprint': 0,
'm': 32,
'factr': 100},
)
self._svm.fit(x, (y + 1) / 2, history)
