TensorFlowClass / pages /17_Graphs2.py
eaglelandsonce's picture
Update pages/17_Graphs2.py
b1de838 verified
raw
history blame
4.69 kB
import streamlit as st
import tensorflow as tf
import tensorflow_gnn as tfgnn
from tensorflow_gnn.models import mt_albis
import networkx as nx
import matplotlib.pyplot as plt
import numpy as np
# Set environment variable for legacy Keras
import os
os.environ['TF_USE_LEGACY_KERAS'] = '1'
# Define the model function
def model_fn(graph_tensor_spec: tfgnn.GraphTensorSpec):
graph = inputs = tf.keras.Input(type_spec=graph_tensor_spec)
# Encode input features to match the required output shape of 128
graph = tfgnn.keras.layers.MapFeatures(
node_sets_fn=lambda node_set, node_set_name: tf.keras.layers.Dense(128)(node_set['features'])
)(graph)
# For each round of message passing...
for _ in range(2):
# ... create and apply a Keras layer.
graph = mt_albis.MtAlbisGraphUpdate(
units=128, message_dim=64,
attention_type="none", simple_conv_reduce_type="mean",
normalization_type="layer", next_state_type="residual",
state_dropout_rate=0.2, l2_regularization=1e-5,
receiver_tag=tfgnn.TARGET # Use TARGET instead of NODES
)(graph)
return tf.keras.Model(inputs, graph)
# Function to create a sample graph with meaningful synthetic data
def create_sample_graph():
num_nodes = 10
num_edges = 15
# Create a random graph
graph = nx.gnm_random_graph(num_nodes, num_edges, directed=True)
# Generate synthetic features
years_published = np.random.randint(1990, 2022, size=num_nodes).astype(np.float32)
num_authors = np.random.randint(1, 10, size=num_nodes).astype(np.float32)
citation_weights = np.random.uniform(0.1, 5.0, size=num_edges).astype(np.float32)
# Combine features into a single array per node
node_features = np.stack([years_published, num_authors], axis=-1)
edge_features = citation_weights.reshape(-1, 1)
# Assign random titles to nodes
paper_titles = [f"Paper {i+1}" for i in range(num_nodes)]
nx.set_node_attributes(graph, {i: {'title': title} for i, title in enumerate(paper_titles)})
graph_tensor = tfgnn.GraphTensor.from_pieces(
node_sets={
"papers": tfgnn.NodeSet.from_fields(
sizes=[num_nodes],
features={"features": tf.convert_to_tensor(node_features)}
)
},
edge_sets={
"cites": tfgnn.EdgeSet.from_fields(
sizes=[num_edges],
adjacency=tfgnn.Adjacency.from_indices(
source=("papers", tf.constant([e[0] for e in graph.edges()], dtype=tf.int32)),
target=("papers", tf.constant([e[1] for e in graph.edges()], dtype=tf.int32))
),
features={"features": tf.convert_to_tensor(edge_features)}
)
}
)
return graph, graph_tensor, node_features, edge_features
# Streamlit app
def main():
st.title("Graph Neural Network Architecture Visualization")
if st.button("Recreate Graph"):
recreate_graph = True
else:
recreate_graph = False
if recreate_graph:
# Create sample graph
nx_graph, graph_tensor, node_features, edge_features = create_sample_graph()
# Create and compile the model
model = model_fn(graph_tensor.spec)
model.compile(optimizer='adam', loss='binary_crossentropy')
# Display model summary
st.subheader("Model Summary")
model.summary(print_fn=lambda x: st.text(x))
# Visualize the graph
st.subheader("Sample Graph Visualization")
fig, ax = plt.subplots(figsize=(10, 8))
pos = nx.spring_layout(nx_graph)
labels = nx.get_node_attributes(nx_graph, 'title')
nx.draw(nx_graph, pos, labels=labels, with_labels=True, node_color='lightblue',
node_size=3000, arrowsize=20, ax=ax) # Increased node_size to 3000
st.pyplot(fig)
# Display graph tensor info
st.subheader("Graph Tensor Information")
st.text(f"Number of nodes: {graph_tensor.node_sets['papers'].total_size}")
st.text(f"Number of edges: {graph_tensor.edge_sets['cites'].total_size}")
st.text(f"Node feature shape: {graph_tensor.node_sets['papers']['features'].shape}")
st.text(f"Edge feature shape: {graph_tensor.edge_sets['cites']['features'].shape}")
# Display sample node and edge features
st.subheader("Sample Node and Edge Features")
st.write("Node Features (Year Published, Number of Authors):")
st.write(node_features)
st.write("Edge Features (Citation Weight):")
st.write(edge_features)
if __name__ == "__main__":
main()