add softmax_edge_neighbors (#59)

* add softmax_edge_neighbors

docs/src/api/utils.md

@@ -25,6 +25,12 @@ GraphNeuralNetworks.broadcast_nodes
 GraphNeuralNetworks.broadcast_edges
 ```
 
+### Neighborhood operations
+
+```@docs
+GraphNeuralNetworks.softmax_edge_neighbors
+```
+
 ### NNlib
 
 Primitive functions implemented in NNlib.jl.

docs/src/messagepassing.md

@@ -14,14 +14,14 @@ A generic message passing on graph takes the form
 where we refer to ``\phi`` as to the message function, 
 and to ``\gamma_x`` and ``\gamma_e`` as to the node update and edge update function
 respectively. The aggregation ``\square`` is over the neighborhood ``N(i)`` of node ``i``, 
-and it is usually set to summation ``\sum``, a max or a mean operation. 
+and it is usually equal either to ``\sum``, to `max` or to a `mean` operation. 
 
 In GNN.jl, the function [`propagate`](@ref) takes care of materializing the
 node features on each edge, applying the message function, performing the
 aggregation, and returning ``\bar{\mathbf{m}}``. 
 It is then left to the user to perform further node and edge updates,
-manypulating arrays of size ``D_{node} \times num_nodes`` and   
-``D_{edge} \times num_edges``.
+manypulating arrays of size ``D_{node} \times num\_nodes`` and   
+``D_{edge} \times num\_edges``.
 
 As part of the [`propagate`](@ref) pipeline, we have the function
 [`apply_edges`](@ref). It can be independently used to materialize 
@@ -34,9 +34,9 @@ and [`NNlib.scatter`](@ref) methods.
 
 ## Examples
 
-### Basic use propagate and apply_edges 
-
+### Basic use of propagate and apply_edges 
 
+TODO
 
 ### Implementing a custom Graph Convolutional Layer
 

src/GraphNeuralNetworks.jl

@@ -35,6 +35,7 @@ export
     reduce_nodes, reduce_edges, 
     softmax_nodes, softmax_edges,
     broadcast_nodes, broadcast_edges,
+    softmax_edge_neighbors,
 
     # msgpass
     apply_edges, propagate,

src/layers/pool.jl

@@ -52,24 +52,24 @@ Global soft attention layer from the [Gated Graph Sequence Neural
 Networks](https://arxiv.org/abs/1511.05493) paper
 
 ```math
-\mathbf{u}_V} = \sum_{i\in V} \alpha_i\, f_{\mathrm{feat}}(\mathbf{x}_i)
+\mathbf{u}_V = \sum_{i\in V} \alpha_i\, f_{feat}(\mathbf{x}_i)
 ```
 
-where the coefficients ``alpha_i`` are given by a [`softmax_nodes`](@ref)
+where the coefficients ``\alpha_i`` are given by a [`softmax_nodes`](@ref)
 operation:
 
 ```math
-\alpha_i = \frac{e^{f_{\mathrm{feat}}(\mathbf{x}_i)}}
-                {\sum_{i'\in V} e^{f_{\mathrm{feat}}(\mathbf{x}_{i'})}}.
+\alpha_i = \frac{e^{f_{gate}(\mathbf{x}_i)}}
+                {\sum_{i'\in V} e^{f_{gate}(\mathbf{x}_{i'})}}.
 ```
 
 # Arguments
 
-- `fgate`: The function ``f_{\mathrm{gate}} \colon \mathbb{R}^{D_{in}} \to
-\mathbb{R}``. It is tipically a neural network.
+- `fgate`: The function ``f_{gate}: \mathbb{R}^{D_{in}} \to \mathbb{R}``. 
+           It is tipically expressed by a neural network.
 
-- `ffeat`: The function ``f_{\mathrm{feat}} \colon \mathbb{R}^{D_{in}} \to
-\mathbb{R}^{D_{out}}``. It is tipically a neural network.
+- `ffeat`: The function ``f_{feat}: \mathbb{R}^{D_{in}} \to \mathbb{R}^{D_{out}}``. 
+           It is tipically expressed by a neural network.
 
 # Examples
 
@@ -88,6 +88,7 @@ g = Flux.batch([GNNGraph(random_regular_graph(10, 4),
 u = pool(g, g.ndata.x)
 
 @assert size(u) == (chout, g.num_graphs)
+```
 """
 struct GlobalAttentionPool{G,F}
     fgate::G

src/utils.jl

@@ -190,6 +190,25 @@ function softmax_edges(g::GNNGraph, e)
     return num ./ den
 end
 
+@doc raw"""
+    softmax_edge_neighbors(g, e)
+
+Softmax over each node's neighborhood of the edge features `e`.
+
+```math
+\mathbf{e}'_{j\to i} = \frac{e^{\mathbf{e}_{j\to i}}}
+                    {\sum_{j'\in N(i)} e^{\mathbf{e}_{j\to i}}}.
+```
+"""
+function softmax_edge_neighbors(g::GNNGraph, e)
+    @assert size(e)[end] == g.num_edges
+    s, t = edge_index(g)
+    max_ = gather(scatter(max, e, t), t)
+    num = exp.(e .- max_)
+    den = gather(scatter(+, num, t), t)
+    return num ./ den
+end
+
 """
     broadcast_nodes(g, x)
 

test/utils.jl

@@ -31,7 +31,6 @@
         @test r[:,1:60] ≈ softmax(getgraph(g, 1).edata.e, dims=2)
     end
 
-
     @testset "broadcast_nodes" begin
         z = rand(4, g.num_graphs)
         r = broadcast_nodes(g, z)
@@ -49,4 +48,15 @@
         @test r[:,60] ≈ z[:,1]
         @test r[:,61] ≈ z[:,2]
     end
+
+    @testset "softmax_edge_neighbors" begin
+        s = [1,2,3,4]
+        t = [5,5,6,6]
+        g2 = GNNGraph(s, t)
+        e2 = randn(Float32, 3, g2.num_edges)
+        z = softmax_edge_neighbors(g2, e2)
+        @test size(z) == size(e2)
+        @test z[:,1:2] ≈ NNlib.softmax(e2[:,1:2], dims=2)
+        @test z[:,3:4] ≈ NNlib.softmax(e2[:,3:4], dims=2)
+    end
 end