Forbu

from torch_geometric.nn.conv import MessagePassing

class MPGNNConv(MessagePassing):
    def __init__(self, node_dim, edge_dim, layers=3):
        super().__init__(aggr='mean', node_dim=0)
        self.lin_edge = MLP(in_dim=node_dim * 2 + edge_dim, out_dim=node_dim, hidden_layers=layers)
        self.lin_node = MLP(in_dim=node_dim * 2, out_dim=node_dim, hidden_layers=layers)

    def forward(self, x, edge_index, edge_attr):
        """

Forbu

Interesting Course :

Introduction to Topological Data Analysis :

Best introduction I'd see so far :

Introduction to Topological Data Analysis

Interesting papers using Topological Data Analysis :

Paper on a very good dimension reduction algorithm :

Forbu

Introduction Material

Course (MOOC)

Portail cours X : http://www.enseignement.polytechnique.fr/informatique/INF556/

Computational Topopology All : https://www2.cs.duke.edu/courses/fall06/cps296.1/

github recap : https://gist.github.com/calstad/01e174faff2cdca7faf9