# Copyright 2022 Huawei Technologies Co., Ltd
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ============================================================================
"""GatedGraphConv Layer"""
import math
import mindspore as ms
from mindspore.common.initializer import XavierUniform
from mindspore_gl import Graph
from .. import GNNCell
class HomoGraphConv(GNNCell):
"""
Homo Graph Conv
Args:
out_feat_size (int): Output node feature size.
bias (bool): Whether use bias.
Returns:
Tensor, output node features.
"""
def __init__(self,
out_feat_size: int,
bias=True):
"""
Init HomoGraphConv.
"""
super().__init__()
gain = math.sqrt(2) # gain for relu
self.dense = ms.nn.Dense(out_feat_size, out_feat_size, has_bias=bias, weight_init=XavierUniform(gain))
# pylint: disable=arguments-differ
def construct(self, x, g: Graph):
"""
Construct function for HomoGraphConv.
"""
g.set_vertex_attr({"h": self.dense(x)})
for v in g.dst_vertex:
v.h = g.sum([u.h for u in v.innbs])
return [v.h for v in g.dst_vertex]
[文档]class GatedGraphConv(ms.nn.Cell):
r"""
Gated Graph Convolution Layer. From the paper `Gated Graph Sequence Neural Networks
<https://arxiv.org/pdf/1511.05493.pdf>`_ .
.. math::
h_{i}^{0} = [ x_i \| \mathbf{0} ] \\
a_{i}^{t} = \sum_{j\in\mathcal{N}(i)} W_{e_{ij}} h_{j}^{t} \\
h_{i}^{t+1} = \mathrm{GRU}(a_{i}^{t}, h_{i}^{t})
Args:
in_feat_size (int): Input node feature size.
out_feat_size (int): Output node feature size.
n_steps (int): Number of steps.
n_etype (int): Number of edge types.
bias (bool, optional): Whether use bias. Default: True.
Inputs:
- **x** (Tensor): The input node features. The shape is :math:`(N,*)` where :math:`N` is the number of nodes,
and :math:`*` could be of any shape.
- **src_idx** (List): The source index for each edge type.
- **dst_idx** (List): The destination index for each edge type.
- **n_nodes** (int): The number of nodes of the whole graph.
- **n_edges** (List): The number of edges for each edge type.
Outputs:
- Tensor, output node features. The shape is :math:`(N, out\_feat\_size)`.
Raises:
TypeError: If `in_feat_size` is not a positive int.
TypeError: If `out_feat_size` is not a positive int.
TypeError: If `n_steps` is not a positive int.
TypeError: If `n_etype` is not a positive int.
TypeError: If `bias` is not a bool.
Supported Platforms:
``Ascend`` ``GPU``
Examples:
>>> import mindspore as ms
>>> from mindspore_gl.nn import GatedGraphConv
>>> from mindspore_gl import GraphField
>>> feat_size = 16
>>> n_nodes = 4
>>> h = ms.ops.Ones()((n_nodes, feat_size), ms.float32)
>>> src_idx = [ms.Tensor([0, 1, 2, 3], ms.int32), ms.Tensor([0, 0, 1, 1], ms.int32),
... ms.Tensor([0, 0, 1, 2, 3], ms.int32), ms.Tensor([2, 3, 3, 0, 1], ms.int32),
... ms.Tensor([0, 1, 2, 3], ms.int32), ms.Tensor([2, 0, 2, 1], ms.int32)]
>>> dst_idx = [ms.Tensor([0, 0, 1, 1], ms.int32), ms.Tensor([0, 1, 2, 3], ms.int32),
... ms.Tensor([2, 3, 3, 0, 1], ms.int32), ms.Tensor([0, 0, 1, 2, 3], ms.int32),
... ms.Tensor([2, 0, 2, 1], ms.int32), ms.Tensor([0, 1, 2, 3], ms.int32)]
>>> n_edges = [4, 4, 5, 5, 4, 4]
>>> conv = GatedGraphConv(feat_size, 16, 2, 6, True)
>>> ret = conv(h, src_idx, dst_idx, n_nodes, n_edges)
>>> print(ret.shape)
(4, 16)
"""
def __init__(self,
in_feat_size: int,
out_feat_size: int,
n_steps: int,
n_etype: int,
bias=True):
super().__init__()
assert isinstance(in_feat_size, int) and in_feat_size > 0, "in_feat_size must be positive int"
assert isinstance(out_feat_size, int) and out_feat_size > 0, "out_feat_size must be positive int"
if in_feat_size > out_feat_size:
raise TypeError("GatedGraphConv requires input feature size <= out_feature_size")
self.in_feat_size = in_feat_size
self.out_feat_size = out_feat_size
cl = []
for _ in range(n_etype):
cl.append(HomoGraphConv(out_feat_size, bias))
self.cell_list = ms.nn.CellList(cl)
self.n_etype = n_etype
self.n_steps = n_steps
self.gru = ms.nn.GRUCell(out_feat_size, out_feat_size)
# pylint: disable=arguments-differ
def construct(self, x, src_idx, dst_idx, n_nodes, n_edges):
"""
Construct function for GatedGraphConv.
"""
if self.in_feat_size < self.out_feat_size:
x = ms.ops.Concat(axis=-1)(x, ms.ops.Zeros()((ms.ops.Shape()(x)[0], self.out_feat_size - self.in_feat_size),
ms.float32))
for _ in range(self.n_steps):
out = self.cell_list[0](x, src_idx[0], dst_idx[0], n_nodes, n_edges)
for i in range(1, self.n_etype):
out += self.cell_list[i](x, src_idx[i], dst_idx[i], n_nodes, n_edges)
if self.gru is not None:
out = self.gru(out, x)
x = out
x = out
return x