galileo_framework_python_base_message_passing_BaseMessagePassing.3.md

date	section	title
2021年 九月 1日 星期三	3	galileo.framework.python.base_message_passing.BaseMessagePassing

NAME

galileo.framework.python.base_message_passing.BaseMessagePassing - Base message passing for GNN

SYNOPSIS

\

继承自 galileo.framework.python.named_object.NamedObject, metaclass , 以及 ABCMeta .

被 galileo.framework.pytorch.python.convolutions.sage_layer.SAGELayer, galileo.framework.pytorch.python.convolutions.sage_layer_sparse.SAGESparseLayer, galileo.framework.tf.python.convolutions.gcn_layer.GCNLayer, galileo.framework.tf.python.convolutions.sage_layer.SAGELayer , 以及 galileo.framework.tf.python.convolutions.sage_layer_sparse.SAGESparseLayer 继承.

Public 成员函数

def __init__ (self, dict config=None, str name=None)\

def config (self)
get config

def __call__ (self, inputs, training=None)\

def message (self, inputs, training=None)
"messagefeaturesonverticesandedges
"

def aggregate (self, inputs)
"aggregatemessagesfromneighborswithtarget
"

def update (self, inputs)
"updatetargetfeatures
"

def message_and_aggregate (self, inputs, training=None)
"messageandaggregatefeaturesonverticesandedges
"

详细描述

Base message passing for GNN

paper 'Neural Message Passing for Quantum Chemistry' <https://arxiv.org/abs/1704.01212>

构造及析构函数说明

def galileo.framework.python.base_message_passing.BaseMessagePassing.__init__ ( self, dict config = None, str name = None)

参数

config dict, config
name name of inputs

成员函数说明

def galileo.framework.python.base_message_passing.BaseMessagePassing.aggregate ( self, inputs)

aggregate messages from neighbors with target
subclass should override this method

参数

inputs inputs for aggregate

返回

tensors

被 galileo.framework.tf.python.convolutions.sage_layer.SAGELayer , 以及 galileo.framework.pytorch.python.convolutions.sage_layer.SAGELayer 重载.

def galileo.framework.python.base_message_passing.BaseMessagePassing.message ( self, inputs, training = None)

message features on vertices and edges
subclass should override this method

参数

inputs inputs for message passing
training

返回

tensors

被 galileo.framework.tf.python.convolutions.sage_layer.SAGELayer , 以及 galileo.framework.pytorch.python.convolutions.sage_layer.SAGELayer 重载.

def galileo.framework.python.base_message_passing.BaseMessagePassing.message_and_aggregate ( self, inputs, training = None)

message and aggregate features on vertices and edges
subclass should override this method

参数

inputs inputs for message passing
training

返回

tensors

被 galileo.framework.tf.python.convolutions.sage_layer_sparse.SAGESparseLayer, galileo.framework.tf.python.convolutions.gcn_layer.GCNLayer , 以及 galileo.framework.pytorch.python.convolutions.sage_layer_sparse.SAGESparseLayer 重载.

def galileo.framework.python.base_message_passing.BaseMessagePassing.update ( self, inputs)

update target features
subclass should override this method

参数

inputs inputs for update

返回

tensors

被 galileo.framework.tf.python.convolutions.sage_layer_sparse.SAGESparseLayer, galileo.framework.tf.python.convolutions.sage_layer.SAGELayer, galileo.framework.tf.python.convolutions.gcn_layer.GCNLayer, galileo.framework.pytorch.python.convolutions.sage_layer_sparse.SAGESparseLayer , 以及 galileo.framework.pytorch.python.convolutions.sage_layer.SAGELayer 重载.

作者

由 Doyxgen 通过分析 Galileo 的 源代码自动生成.