mindspore_gl.nn.GatedGraphConv

class mindspore_gl.nn.GatedGraphConv(in_feat_size: int, out_feat_size: int, n_steps: int, n_etype: int, bias=True)[source]

Gated Graph Convolution Layer. From the paper Gated Graph Sequence Neural Networks .

\[ \begin{align}\begin{aligned}\begin{split}h_{i}^{0} = [ x_i \| \mathbf{0} ] \\\end{split}\\\begin{split}a_{i}^{t} = \sum_{j\in\mathcal{N}(i)} W_{e_{ij}} h_{j}^{t} \\\end{split}\\h_{i}^{t+1} = \mathrm{GRU}(a_{i}^{t}, h_{i}^{t})\end{aligned}\end{align} \]
Parameters

  • 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) – Whether use bias. Default: True.

Inputs:

  • x (Tensor): The input node features. The shape is \((N,*)\) where \(N\) is the number of nodes, and \(*\) 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 \((N, out\_feat\_size)\).

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)