mindspore_gl.nn.GINConv

class mindspore_gl.nn.GINConv(activation, init_eps=0.0, learn_eps=False, aggregation_type='sum')[source]

Graph isomorphic network layer. From the paper How Powerful are Graph Neural Networks? .

\[h_i^{(l+1)} = f_\Theta \left((1 + \epsilon) h_i^{l} + \mathrm{aggregate}\left(\left\{h_j^{l}, j\in\mathcal{N}(i) \right\}\right)\right)\]

If weights are provided on each edge, the weighted graph convolution is defined as:

\[h_i^{(l+1)} = f_\Theta \left((1 + \epsilon) h_i^{l} + \mathrm{aggregate}\left(\left\{e_{ji} h_j^{l}, j\in\mathcal{N}(i) \right\}\right)\right)\]
Parameters

  • activation (mindspore.nn.Cell) – Activation function.

  • init_eps (float, optional) – Init value of eps. Default: 0.

  • learn_eps (bool, optional) – Whether eps is learnable. Default: False.

  • aggregation_type (str, optional) – Type of aggregation, should in ‘sum’, ‘max’ and ‘avg’. Default: ‘sum’.

Inputs:

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

  • edge_weight (Tensor): The input edge weights. The shape is \((M,*)\) where \(M\) is the number of nodes, and \(*\) could be of any shape.

  • g (Graph): The input graph.

Outputs:

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

Raises

  • TypeError – If activation is not a mindspore.nn.Cell.

  • TypeError – If init_eps is not a float.

  • TypeError – If learn_eps is not a bool.

  • SyntaxError – Raised when the aggregation_type not in ‘sum’, ‘max’ and ‘avg’.

Supported Platforms:

Ascend GPU

Examples

>>> import mindspore as ms
>>> from mindspore_gl.nn import GINConv
>>> 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)
>>> edges_weight = ones((n_edges, feat_size), ms.float32)
>>> graph_field = GraphField(src_idx, dst_idx, n_nodes, n_edges)
>>> conv = GINConv(activation=None, init_eps=0., learn_eps=False, aggregation_type="sum")
>>> ret = conv(nodes_feat, edges_weight, *graph_field.get_graph())
>>> print(ret.shape)
(4, 16)