# import keras_core as ks
from ._model import model_disjoint, model_disjoint_edge
from kgcnn.models.utils import update_model_kwargs
from kgcnn.layers.scale import get as get_scaler
from kgcnn.layers.modules import Input
from kgcnn.models.casting import (template_cast_output, template_cast_list_input,
template_cast_list_input_docs, template_cast_output_docs)
from keras.backend import backend as backend_to_use
from kgcnn.ops.activ import *
# Keep track of model version from commit date in literature.
# To be updated if model is changed in a significant way.
__model_version__ = "2023-09-18"
# Supported backends
__kgcnn_model_backend_supported__ = ["tensorflow", "torch", "jax"]
if backend_to_use() not in __kgcnn_model_backend_supported__:
raise NotImplementedError("Backend '%s' for model 'GIN' is not supported." % backend_to_use())
# Implementation of GIN in `keras` from paper:
# How Powerful are Graph Neural Networks?
# Keyulu Xu, Weihua Hu, Jure Leskovec, Stefanie Jegelka
# https://arxiv.org/abs/1810.00826
model_default = {
"name": "GIN",
"inputs": [
{"shape": (None,), "name": "node_number", "dtype": "int64"},
{"shape": (None, 2), "name": "edge_indices", "dtype": "int64"},
{"shape": (), "name": "total_nodes", "dtype": "int64"},
{"shape": (), "name": "total_edges", "dtype": "int64"}
],
"input_tensor_type": "padded",
"cast_disjoint_kwargs": {},
"input_embedding": None, # deprecated
"input_node_embedding": {"input_dim": 95, "output_dim": 64},
"gin_mlp": {"units": [64, 64], "use_bias": True, "activation": ["relu", "linear"],
"use_normalization": True, "normalization_technique": "graph_batch",
"padded_disjoint": False},
"gin_args": {},
"depth": 3, "dropout": 0.0, "verbose": 10,
"last_mlp": {"use_bias": [True, True, True], "units": [64, 64, 64],
"activation": ["relu", "relu", "linear"]},
"output_embedding": 'graph',
"output_to_tensor": None, # deprecated
"output_tensor_type": "padded",
"output_mlp": {"use_bias": True, "units": 1,
"activation": "softmax"},
"output_scaling": None,
}
[docs]@update_model_kwargs(model_default, update_recursive=0, deprecated=["input_embedding", "output_to_tensor"])
def make_model(inputs: list = None,
input_tensor_type: str = None,
cast_disjoint_kwargs: dict = None,
input_embedding: dict = None, # noqa
input_node_embedding: dict = None,
depth: int = None,
gin_args: dict = None,
gin_mlp: dict = None,
last_mlp: dict = None,
dropout: float = None,
name: str = None,
verbose: int = None, # noqa
output_embedding: str = None,
output_to_tensor: bool = None, # noqa
output_mlp: dict = None,
output_scaling: dict = None,
output_tensor_type: str = None,
):
r"""Make `GIN <https://arxiv.org/abs/1810.00826>`__ graph network via functional API.
Default parameters can be found in :obj:`kgcnn.literature.GIN.model_default`.
**Model inputs**:
Model uses the list template of inputs and standard output template.
The supported inputs are :obj:`[nodes, edge_indices, ...]`
with '...' indicating mask or ID tensors following the template below:
%s
**Model outputs**:
The standard output template:
%s
Args:
inputs (list): List of dictionaries unpacked in :obj:`tf.keras.layers.Input`. Order must match model definition.
input_tensor_type (str): Input type of graph tensor. Default is "padded".
cast_disjoint_kwargs (dict): Dictionary of arguments for castin layers.
input_embedding (dict): Deprecated in favour of input_node_embedding etc.
input_node_embedding (dict): Dictionary of arguments for nodes unpacked in :obj:`Embedding` layers.
depth (int): Number of graph embedding units or depth of the network.
gin_args (dict): Dictionary of layer arguments unpacked in :obj:`GIN` convolutional layer.
gin_mlp (dict): Dictionary of layer arguments unpacked in :obj:`MLP` for convolutional layer.
last_mlp (dict): Dictionary of layer arguments unpacked in last :obj:`MLP` layer before output or pooling.
dropout (float): Dropout to use.
name (str): Name of the model.
verbose (int): Level of print output.
output_embedding (str): Main embedding task for graph network. Either "node", "edge" or "graph".
output_to_tensor (bool): Deprecated in favour of `output_tensor_type` .
output_mlp (dict): Dictionary of layer arguments unpacked in the final classification :obj:`MLP` layer block.
Defines number of model outputs and activation.
output_scaling (dict): Dictionary of layer arguments unpacked in scaling layers. Default is None.
output_tensor_type (str): Output type of graph tensors such as nodes or edges. Default is "padded".
Returns:
:obj:`keras.models.Model`
"""
# Make input
model_inputs = [Input(**x) for x in inputs]
disjoint_inputs = template_cast_list_input(
model_inputs, input_tensor_type=input_tensor_type, cast_disjoint_kwargs=cast_disjoint_kwargs,
mask_assignment=[0, 1],
index_assignment=[None, 0]
)
n, disjoint_indices, batch_id_node, batch_id_edge, node_id, edge_id, count_nodes, count_edges = disjoint_inputs
# Wrapping disjoint model.
out = model_disjoint(
[n, disjoint_indices, batch_id_node, count_nodes],
use_node_embedding=("int" in inputs[0]['dtype']) if input_node_embedding is not None else False,
input_node_embedding=input_node_embedding,
depth=depth, gin_args=gin_args, gin_mlp=gin_mlp,
last_mlp=last_mlp, dropout=dropout,
output_embedding=output_embedding, output_mlp=output_mlp
)
if output_scaling is not None:
scaler = get_scaler(output_scaling["name"])(**output_scaling)
out = scaler(out)
# Output embedding choice
out = template_cast_output(
[out, batch_id_node, batch_id_edge, node_id, edge_id, count_nodes, count_edges],
output_embedding=output_embedding, output_tensor_type=output_tensor_type,
input_tensor_type=input_tensor_type, cast_disjoint_kwargs=cast_disjoint_kwargs
)
model = ks.models.Model(inputs=model_inputs, outputs=out, name=name)
model.__kgcnn_model_version__ = __model_version__
if output_scaling is not None:
def set_scale(*args, **kwargs):
scaler.set_scale(*args, **kwargs)
setattr(model, "set_scale", set_scale)
return model
make_model.__doc__ = make_model.__doc__ % (template_cast_list_input_docs, template_cast_output_docs)
model_default_edge = {
"name": "GINE",
"inputs": [
{"shape": (None,), "name": "node_number", "dtype": "int64"},
{"shape": (None,), "name": "edge_number", "dtype": "int64"},
{"shape": (None, 2), "name": "edge_indices", "dtype": "int64"},
{"shape": (), "name": "total_nodes", "dtype": "int64"},
{"shape": (), "name": "total_edges", "dtype": "int64"}
],
"input_tensor_type": "padded",
"cast_disjoint_kwargs": {},
"input_embedding": None, # deprecated
"input_node_embedding": {"input_dim": 95, "output_dim": 64},
"input_edge_embedding": {"input_dim": 10, "output_dim": 64},
"gin_mlp": {"units": [64, 64], "use_bias": True, "activation": ["relu", "linear"],
"use_normalization": True, "normalization_technique": "graph_batch"},
"gin_args": {"epsilon_learnable": False},
"depth": 3, "dropout": 0.0, "verbose": 10,
"last_mlp": {"use_bias": [True, True, True], "units": [64, 64, 64],
"activation": ["relu", "relu", "linear"]},
"output_embedding": 'graph',
"output_to_tensor": None,
"output_scaling": None,
"output_tensor_type": "padded",
"output_mlp": {"use_bias": True, "units": 1,
"activation": "softmax"}
}
[docs]@update_model_kwargs(model_default_edge, update_recursive=0, deprecated=["input_embedding", "output_to_tensor"])
def make_model_edge(inputs: list = None,
input_tensor_type: str = None,
cast_disjoint_kwargs: dict = None,
input_embedding: dict = None, # noqa
input_node_embedding: dict = None,
input_edge_embedding: dict = None,
depth: int = None,
gin_args: dict = None,
gin_mlp: dict = None,
last_mlp: dict = None,
dropout: float = None,
name: str = None,
verbose: int = None, # noqa
output_embedding: str = None,
output_to_tensor: bool = None, # noqa
output_mlp: dict = None,
output_scaling: dict = None,
output_tensor_type: str = None,
):
r"""Make `GINE <https://arxiv.org/abs/1905.12265>`__ graph network via functional API.
Default parameters can be found in :obj:`kgcnn.literature.GIN.model_default_edge` .
Model inputs:
Model uses the list template of inputs and standard output template.
The supported inputs are :obj:`[nodes, edges, edge_indices, ...]`
with '...' indicating mask or ID tensors following the template below:
%s
Model outputs:
The standard output template:
%s
Args:
inputs (list): List of dictionaries unpacked in :obj:`Input`. Order must match model definition.
input_tensor_type (str): Input type of graph tensor. Default is "padded".
cast_disjoint_kwargs (dict): Dictionary of arguments for casting layers.
input_embedding (dict): Deprecated in favour of input_node_embedding etc.
input_node_embedding (dict): Dictionary of arguments for nodes unpacked in :obj:`Embedding` layers.
input_edge_embedding (dict): Dictionary of arguments for edge unpacked in :obj:`Embedding` layers.
depth (int): Number of graph embedding units or depth of the network.
gin_args (dict): Dictionary of layer arguments unpacked in :obj:`GIN` convolutional layer.
gin_mlp (dict): Dictionary of layer arguments unpacked in :obj:`MLP` for convolutional layer.
last_mlp (dict): Dictionary of layer arguments unpacked in last :obj:`MLP` layer before output or pooling.
dropout (float): Dropout to use.
name (str): Name of the model.
verbose (int): Level of print output.
output_embedding (str): Main embedding task for graph network. Either "node", "edge" or "graph".
output_to_tensor (bool): Deprecated in favour of `output_tensor_type` .
output_mlp (dict): Dictionary of layer arguments unpacked in the final classification :obj:`MLP` layer block.
Defines number of model outputs and activation.
output_scaling (dict): Dictionary of layer arguments unpacked in scaling layers. Default is None.
output_tensor_type (str): Output type of graph tensors such as nodes or edges. Default is "padded".
Returns:
:obj:`keras.models.Model`
"""
# Make input
model_inputs = [Input(**x) for x in inputs]
disjoint_inputs = template_cast_list_input(
model_inputs,
input_tensor_type=input_tensor_type,
cast_disjoint_kwargs=cast_disjoint_kwargs,
mask_assignment=[0, 1, 1],
index_assignment=[None, None, 0]
)
n, ed, disjoint_indices, batch_id_node, batch_id_edge, node_id, edge_id, count_nodes, count_edges = disjoint_inputs
# Wrapping disjoint model.
out = model_disjoint_edge(
[n, ed, disjoint_indices, batch_id_node, count_nodes],
use_node_embedding=("int" in inputs[0]['dtype']) if input_node_embedding is not None else False,
use_edge_embedding=("int" in inputs[1]['dtype']) if input_edge_embedding is not None else False,
input_node_embedding=input_node_embedding,
input_edge_embedding=input_edge_embedding,
depth=depth, gin_args=gin_args, gin_mlp=gin_mlp, last_mlp=last_mlp,
dropout=dropout, output_embedding=output_embedding, output_mlp=output_mlp
)
if output_scaling is not None:
scaler = get_scaler(output_scaling["name"])(**output_scaling)
out = scaler(out)
# Output embedding choice
out = template_cast_output(
[out, batch_id_node, batch_id_edge, node_id, edge_id, count_nodes, count_edges],
output_embedding=output_embedding, output_tensor_type=output_tensor_type,
input_tensor_type=input_tensor_type, cast_disjoint_kwargs=cast_disjoint_kwargs
)
model = ks.models.Model(inputs=model_inputs, outputs=out, name=name)
model.__kgcnn_model_version__ = __model_version__
if output_scaling is not None:
def set_scale(*args, **kwargs):
scaler.set_scale(*args, **kwargs)
setattr(model, "set_scale", set_scale)
return model
make_model_edge.__doc__ = make_model_edge.__doc__ % (template_cast_list_input_docs, template_cast_output_docs)