GraphiT is an instance of transformers designed for graph-structured data. It takes as input a graph seen as a set of its node features, and incorporate the graph structure via i) relative positional encoding using kernels on graphs and ii) encoding local substructures around each node, such as short paths, before adding it to the node features.
Optimal transport kernel (OTK) is a kernel for feature aggregation. It allows to perform adaptive pooling (attention + pooling). Principally, it can be useful to model any data represented as sets of features (sequences, images, graphs etc.). In this implementation, it can be used as a module in neural networks, or alone as a kernel method.
Graph convolutional kernel networks (GCKN) is a software package to model graph-structured data.
Recurrent kernel networks (RKN) is a software package to model biological sequences (DNA, protein etc.) with potentially gapped motifs.
CKN-Pytorch-image is a software package to perform image classification with convolutional kernel networks.
CKN-seq is a software package to model biological sequences (DNA, protein etc.) with convolutional kernel networks.