medgeconv 0.2.1

No project description provided

An efficient tensorflow 2 implementation of the edge-convolution layer EdgeConv used in e.g. ParticleNet.

The structure of the layer is as described in ‘ParticleNet: Jet Tagging via Particle Clouds’ https://arxiv.org/abs/1902.08570. Graphs often have a varying number of nodes. By making use of the disjoint unioon of graphs in a batch, memory intensive operations in this implementation are done only on the actual nodes. This is faster if the number of nodes varies between graphs in the batch.

Install e.g. via:

pip install git+https://github.com/StefReck/MEdgeConv.git#egg=MEdgeConv

Use e.g. like this:

import medgeconv

nodes = medgeconv.DisjointEdgeConvBlock(
    units=[64, 64, 64],
    next_neighbors=16,
    to_disjoint=True,
    pooling=True)((nodes, is_valid, coordinates))

Inputs to EdgeConv are 3 dense tensors: nodes, is_valid and coordinates

• nodes, shape (batchsize, n_nodes_max, n_features)

  Node features of the graph, padded to fixed size.

• is_valid, shape (batchsize, n_nodes_max)

  1 for actual node, 0 for padded node.

• coordinates, shape (batchsize, n_nodes_max, n_coords)

  Features of each node used for calculating nearest neighbors.

By using to_disjoint = True, the dense tensors get transformed to the disjoint union. The output is also disjoint. pooling = True will attach a node-wise global average pooling layer in the end.

A full model could look like this:

import tensorflow as tf
import medgeconv

inp = (nodes, is_valid, coordinates)
x = medgeconv.DisjointEdgeConvBlock(
    units=[64, 64, 64],
    to_disjoint=True,
    batchnorm_for_nodes=True)(inp)

x = medgeconv.DisjointEdgeConvBlock(
    units=[128, 128, 128])(x)

x = medgeconv.DisjointEdgeConvBlock(
    units=[256, 256, 256],
    pooling=True)(x)

output = tf.keras.layers.Dense(2)(x)
model = tf.keras.Model(inp, output)

To load models, use the custom_objects:

import medgeconv

model = load_model(path, custom_objects=medgeconv.custom_objects)

Remarks:

• Batchsize has to be fixed (i.e. use Input(batch_size=bs, …))

• in nodes array, valid nodes have to come first, then the padded nodes