mindspore_gl.nn.DOTGATConv

class mindspore_gl.nn.DOTGATConv(in_feat_size: int, out_feat_size: int, num_heads: int, bias=False)[source]

Applying a dot product version of self-attention in GAT. From the paper Graph Attention Network .

\[h_i^{(l+1)} = \sum_{j\in \mathcal{N}(i)} \alpha_{i, j} h_j^{(l)}\]

\(\alpha_{i, j}\) represents the attention score between node \(i\) and node \(j\).

\[\begin{split}\alpha_{i, j} = \mathrm{softmax_i}(e_{ij}^{l}) \\ e_{ij}^{l} = ({W_i^{(l)} h_i^{(l)}})^T \cdot {W_j^{(l)} h_j^{(l)}}\end{split}\]
Parameters

  • in_feat_size (int) – Input node feature size.

  • out_feat_size (int) – Output node feature size.

  • num_heads (int) – Number of attention head used in GAT.

  • bias (bool, optional) – Whether use bias. Default: False.

Inputs:

  • x (Tensor): The input node features. The shape is \((N,*)\) where \(N\) is the number of nodes, and \(*\) could be of any shape.

  • g (Graph): The input graph.

Outputs:

  • Tensor, output node features. The shape is \((N, num\_heads, out\_feat\_size)\).

Raises

  • TypeError – If ‘in_feat_size’ is not a positive int.

  • TypeError – If ‘out_feat_size’ is not a positive int.

  • TypeError – If ‘num_heads’ 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 DOTGATConv
>>> from mindspore_gl import GraphField
>>> n_nodes = 4
>>> n_edges = 8
>>> feat_size = 16
>>> src_idx = ms.Tensor([0, 0, 0, 1, 1, 1, 2, 3], ms.int32)
>>> dst_idx = ms.Tensor([0, 1, 3, 1, 2, 3, 3, 2], ms.int32)
>>> ones = ms.ops.Ones()
>>> nodes_feat = ones((n_nodes, feat_size), ms.float32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
>>> out_size = 4
>>> conv = DOTGATConv(feat_size, out_size, num_heads=2, bias=True)
>>> ret = conv(nodes_feat, *graph_field.get_graph())
>>> print(ret.shape)
(4, 2, 4)