def build(self,
hiddens=[16],
activations=['relu'],
dropout=0.5,
lr=0.01,
weight_decay=5e-4,
use_bias=False,
p1=1.,
p2=1.,
n_power_iterations=1,
epsilon=0.03,
xi=1e-6):
if self.backend == "tensorflow":
with tf.device(self.device):
self.model = tfGCN(self.graph.num_node_attrs,
self.graph.num_node_classes,
hiddens=hiddens,
activations=activations,
dropout=dropout,
weight_decay=weight_decay,
lr=lr,
use_bias=use_bias)
self.index_all = tf.range(self.graph.num_nodes,
dtype=self.intx)
else:
raise NotImplementedError
self.p1 = p1 # Alpha
self.p2 = p2 # Beta
self.xi = xi # Small constant for finite difference
# Norm length for (virtual) adversarial training
self.epsilon = epsilon
self.n_power_iterations = n_power_iterations # Number of power iterations
def build(self,
hiddens=[16],
activations=['relu'],
dropout=0.5,
weight_decay=5e-4,
lr=0.01,
use_bias=False):
if self.backend == "tensorflow":
with tf.device(self.device):
self.model = tfGCN(self.graph.num_node_attrs,
self.graph.num_node_classes,
hiddens=hiddens,
activations=activations,
dropout=dropout,
weight_decay=weight_decay,
lr=lr,
use_bias=use_bias)
else:
self.model = pyGCN(self.graph.num_node_attrs,
self.graph.num_node_classes,
hiddens=hiddens,
activations=activations,
dropout=dropout,
weight_decay=weight_decay,
lr=lr,
use_bias=use_bias).to(self.device)
def build(self,
hiddens=[16],
activations=['relu'],
dropout=0.5,
lr=0.01,
weight_decay=5e-4,
use_bias=False,
p1=1.,
p2=1.,
n_power_iterations=1,
epsilon=0.03,
xi=1e-6):
with tf.device(self.device):
self.model = tfGCN(self.graph.num_node_attrs,
self.graph.num_node_classes,
hiddens=hiddens,
activations=activations,
dropout=dropout,
weight_decay=weight_decay,
lr=lr,
use_bias=use_bias)
self.register_cache(
"index_all", tf.range(self.graph.num_nodes, dtype=self.intx))
self.register_cache("p1", p1) # Alpha
self.register_cache("p2", p2) # Beta
self.register_cache("xi", xi) # Small constant for finite difference
# Norm length for (virtual) adversarial training
self.register_cache("epsilon", epsilon)
self.register_cache("n_power_iterations",
n_power_iterations) # Number of power iterations
def build(self,
hiddens=[32],
activations=['relu'],
dropout=0.5,
weight_decay=0.,
lr=0.01,
use_bias=False):
with tf.device(self.device):
self.model = tfGCN(self.graph.num_node_attrs,
self.graph.num_node_classes,
hiddens=hiddens,
activations=activations,
dropout=dropout,
weight_decay=weight_decay,
lr=lr,
use_bias=use_bias,
experimental_run_tf_function=False)