def test_call(self, step, res):
"""Test call.
Test that attribute of internal optimizer is correctly rerouted to the
internal optimizer
Args:
step: step number to 'GammaBetaDecreasingStep' 'Scheduler'.
res: expected result from call to 'GammaBetaDecreasingStep' 'Scheduler'.
"""
beta = _ops.convert_to_tensor_v2(2, dtype=tf.float32)
gamma = _ops.convert_to_tensor_v2(1, dtype=tf.float32)
scheduler = opt.GammaBetaDecreasingStep()
scheduler.initialize(beta, gamma)
step = _ops.convert_to_tensor_v2(step, dtype=tf.float32)
lr = scheduler(step)
self.assertAllClose(lr.numpy(), res)
def radius(self):
"""Radius, R, of the hypothesis space W.
W is a convex set that forms the hypothesis space.
Returns:
a tensor
"""
return _ops.convert_to_tensor_v2(self.radius_constant, dtype=tf.float32)
def lipchitz_constant(self, class_weight): # pylint: disable=unused-argument
"""Lipchitz constant, L.
Args:
class_weight: class weights used
Returns:
constant L
"""
return _ops.convert_to_tensor_v2(1, dtype=tf.float32)
def beta(self, class_weight): # pylint: disable=unused-argument
"""Smoothness, beta.
Args:
class_weight: the class weights as scalar or 1d tensor, where its
dimensionality is equal to the number of outputs.
Returns:
Beta
"""
return _ops.convert_to_tensor_v2(1, dtype=tf.float32)
def test_run():
loss = TestLoss(1, 1, 1)
bolton = opt.BoltOn(TestOptimizer(), loss)
model = TestModel(1, (1,), 1)
model.compile(bolton, loss)
model.layers[0].kernel = \
model.layers[0].kernel_initializer((model.layer_input_shape[0],
model.n_outputs))
with bolton(noise, epsilon, model.layers, class_weights, 1, 1) as _:
pass
return _ops.convert_to_tensor_v2(bolton.epsilon, dtype=tf.float32)
def project_fn(r):
loss = TestLoss(1, 1, r)
bolton = opt.BoltOn(TestOptimizer(), loss)
model = TestModel(n_out, shape, init_value)
model.compile(bolton, loss)
model.layers[0].kernel = \
model.layers[0].kernel_initializer((model.layer_input_shape[0],
model.n_outputs))
bolton._is_init = True # pylint: disable=protected-access
bolton.layers = model.layers
bolton.epsilon = 2
bolton.noise_distribution = 'laplace'
bolton.n_outputs = 1
bolton.n_samples = 1
bolton.project_weights_to_r()
return _ops.convert_to_tensor_v2(bolton.layers[0].kernel, tf.float32)
def test_run(fn, args):
loss = TestLoss(1, 1, 1)
bolton = opt.BoltOn(TestOptimizer(), loss)
model = TestModel(1, (1, ), 1)
model.compile(bolton, loss)
model.layers[0].kernel = \
model.layers[0].kernel_initializer((model.layer_input_shape[0],
model.n_outputs))
bolton._is_init = True # pylint: disable=protected-access
bolton.noise_distribution = 'laplace'
bolton.epsilon = 1
bolton.layers = model.layers
bolton.class_weights = 1
bolton.n_samples = 1
bolton.batch_size = 1
bolton.n_outputs = 1
res = getattr(bolton, fn, lambda: 'test')(*args)
if res != 'test':
res = 1
else:
res = 0
return _ops.convert_to_tensor_v2(res, dtype=tf.float32)
def gamma(self):
"""Returns strongly convex parameter, gamma."""
return _ops.convert_to_tensor_v2(1, dtype=tf.float32)