torch_geometric.nn.models.MetaPath2Vec
- class MetaPath2Vec(edge_index_dict: Dict[Tuple[str, str, str], Tensor], embedding_dim: int, metapath: List[Tuple[str, str, str]], walk_length: int, context_size: int, walks_per_node: int = 1, num_negative_samples: int = 1, num_nodes_dict: Optional[Dict[str, int]] = None, sparse: bool = False)[source]
Bases:
Module来自“metapath2vec: 异构网络的可扩展表示学习”论文的MetaPath2Vec模型，其中基于给定的
metapath在异构图中进行随机游走采样，并通过负采样优化学习节点嵌入。注意
有关使用MetaPath2Vec的示例，请参见 examples/hetero/metapath2vec.py。
- Parameters:
edge_index_dict (Dict[Tuple[str, str, str], torch.Tensor]) – 字典 保存每个
(src_node_type, rel_type, dst_node_type)边类型的边索引 在异质图中。embedding_dim (int) – The size of each embedding vector.
metapath (List[Tuple[str, str, str]]) – 元路径描述为一个由
(src_node_type, rel_type, dst_node_type)元组组成的列表。walk_length (int) – The walk length.
context_size (int) – The actual context size which is considered for positive samples. This parameter increases the effective sampling rate by reusing samples across different source nodes.
walks_per_node (int, optional) – The number of walks to sample for each node. (default:
1)num_negative_samples (int, optional) – The number of negative samples to use for each positive sample. (default:
1)num_nodes_dict (Dict[str, int], optional) – 保存每种节点类型的节点数量的字典。（默认值：
None）sparse (bool, optional) – If set to
True, gradients w.r.t. to the weight matrix will be sparse. (default:False)
- forward(node_type: str, batch: Optional[Tensor] = None) Tensor[source]
返回类型为
node_type的batch中节点的嵌入。- Return type:
- loader(**kwargs)[source]
返回在异质图上创建正负随机游走的数据加载器。
- Parameters:
**kwargs (可选) –
torch.utils.data.DataLoader的参数，例如batch_size,shuffle,drop_last或num_workers.