Update gsoc.md

docs/src/gsoc.md

@@ -1,36 +1,3 @@
 # Graph Neural Networks - Summer of Code
 
-Graph neural networks (GNN) are deep learning models well adapated to data  rapresented by graphs along with feature vectors associated to nodes and edges. 
-GNNs are a growing are of research and find many applications in complex networks analysis, relational reasoning, combinatorial optimization, molecule generation, and many other fields. 
-[GraphNeuralNetworks.jl](https://github.com/CarloLucibello/GraphNeuralNetworks.jl) is a pure Julia package for GNNs already equipped with plenty of features such as many common graph convolutional layers, CUDA support and graph batching for fast parallel operations. There are a number of ways by which the package could be improved:
-
-- **Adding graph convolutional layers (duration: 175h, expected difficulty: easy to medium):**
-While we implement a good variety of graph convolutional layers, there is still a vast zoology to be implemented yet.
-
-- **Adding models and examples (duration: 175h, expected difficulty: medium):**
-As part of the documentation and for bootstrapping new projects, 
-we want to add fully worked out examples and applications of graph neural networks.
-
-- **Adding graph datasets (duration: 175h, expected difficulty: easy):**
-. Provide julia friendly wrappers for common graph datasets in [`MLDatasets.jl`](https://github.com/JuliaML/MLDatasets.jl).
-
-- **Supporting heterogeneous graphs (duration: 175h, expected difficulty: hard):**
-In some complex networks, the relations expressed by edges can be of different types.
-We need to implement an heteroeneous graph type and implement convolutional layers supporting them. 
-
-- **Training on very large graphs (duration: 175h, expected difficulty: medium to hard):**
-Graph containing several milions of nodes are too large for gpu memory. Mini-batch training si performed on subgraphs, as in the GraphSAGE algorithm.
-
-- **Supporting temporal graph neural networks (duration: 175h, expected difficulty: hard):**
-We aim at implementing temporal graph convolutions for time-varying graph and/or node features.
-
-- **Improving perfomance using sparse linear algebra (duration: 175h, expected difficulty: medium to hard):** 
-Many graph convolutional layers can be expressed as non-materializing algebraic operations involving the adjacency matrix instead of the slower and more memory consuming gather/scatter mechanism. We aim at extending as far as possible and in a gpu-friendly way these *fused* implementation.
-
-**Recommended skills:** Familiarity with graph neural networks and Flux.jl.
-
-**Expected results:** New features added to the package along with tests and relevant documentation.
-
-**Mentors:** [Carlo Lucibello](https://github.com/CarloLucibello) (author of [GraphNeuralNetworks.jl](https://github.com/CarloLucibello/GraphNeuralNetworks.jl).For linear algebra, co-mentoring by [Will Kimmerer](https://github.com/Wimmerer) (lead developer of [SuiteSparseGraphBLAS.jl](https://github.com/JuliaSparse/SuiteSparseGraphBLAS.jl)).
-
-**Contact:** Feel free to contact us on the [Julia Slack Workspace](https://julialang.slack.com/) or by opening an issue on [GraphNeuralNetworks.jl](https://github.com/CarloLucibello/GraphNeuralNetworks.jl).
+Potential candidates to Google Summer of Code's scholarships can find out about the available projects involving GraphNeuralNetworks.jl on the [dedicated page](https://julialang.org/jsoc/gsoc/gnn/) in the Julia Language website.