Flexible edge-empowered graph convolutional networks with node-edge enhancement
Systems and methods for predicting road conditions and traffic volume is provided. The method includes generating a graph of one or more road regions including a plurality of road intersections and a plurality of road segments, wherein the road intersections are represented as nodes and the road segments are represented as edges. The method can also include embedding the nodes from the graph into a node space, translating the edges of the graph into nodes of a line graph, and embedding the nodes of the line graph into the node space. The method can also include aligning the nodes from the line graph with the nodes from the graph, and optimizing the alignment, outputting a set of node and edge representations that predicts the traffic flow for each of the road segments and road intersections based on the optimized alignment of the nodes.
1. A computer implemented method for predicting road conditions and traffic volume, comprising:
generating a graph of one or more road regions, wherein each road region includes a plurality of road intersections and a plurality of road segments, and wherein the road intersections are represented as nodes in the graph and the road segments are represented as edges in the graph;
embedding the nodes from the graph into a node space;
translating the edges of the graph into nodes of a line graph;
embedding the nodes of the line graph into the node space;
aligning the nodes from the line graph with the nodes from the graph using a learned mapping matrix;
optimizing the alignment by utilizing a regularization term in a final loss function; and
outputting a set of node and edge representations that predicts the traffic flow for each of the road segments and road intersections of the graph based on the optimized alignment of the nodes from the line graph with the nodes from the graph.
2. The computer implemented method of claim 1 , further comprising directing road improvements for the road segments and road intersections predicted to have the greatest traffic flows.
3. The computer implemented method of claim 1 , further comprising updating the edge embeddings and node embedding by modeling meta-interactions between the nodes and edges of the graph.
4. The computer implemented method of claim 3 , wherein the meta-interactions include node-to-edge interactions, edge-to-node interactions, and node-to-node interactions.
5. The computer implemented method of claim 1 , wherein the graph is an attributed graphs represented as G=(V, E, X v , X E ), where V denotes the set of nodes, E denotes the set of edges, X v denotes the set of node attributes, and X E denotes the set of edge attributes.
6. The computer implemented method of claim 5 , wherein the edge attributes are only partially available.
7. The computer implemented method of claim 6 , wherein the edge attributes are selected from the group consisting of traffic volume, road conditions, hazards, flooding, number of lanes, and combinations thereof.
8. A non-transitory computer readable storage medium comprising a computer readable program for predicting road conditions and traffic volume, wherein the computer readable program when executed on a computer causes the computer to perform the steps of:
generating a graph of one or more road regions, wherein each road region includes a plurality of road intersections and a plurality of road segments, and wherein the road intersections are represented as nodes in the graph and the road segments are represented as edges in the graph;
embedding the nodes from the graph into a node space;
translating the edges of the graph into nodes of a line graph;
embedding the nodes of the line graph into the node space;
aligning the nodes from the line graph with the nodes from the graph using a learned mapping matrix;
optimizing the alignment by utilizing a regularization term in a final loss function; and
outputting a set of node and edge representations that predicts the traffic flow for each of the road segments and road intersections of the graph based on the optimized alignment of the nodes from the line graph with the nodes from the graph.
9. The computer readable program as recited in claim 8 , further comprising directing road improvements for the road segments and road intersections predicted to have the greatest traffic flows.
10. The computer readable program as recited in claim 8 , further comprising updating the edge embeddings and node embedding by modeling meta-interactions between the nodes and edges of the graph.
11. The computer readable program as recited in claim 10 , wherein the meta-interactions include node-to-edge interactions, edge-to-node interactions, and node-to-node interactions.
12. The computer readable program as recited in claim 8 , wherein the graph is an attributed graphs represented as G=(V, E, X v , X E ), where V denotes the set of nodes, E denotes the set of edges, X v denotes the set of node attributes, and X E denotes the set of edge attributes.
13. The computer readable program as recited in claim 12 , wherein the edge attributes are only partially available.
14. The computer readable program as recited in claim 13 , wherein the edge attributes are selected from the group consisting of traffic volume, road conditions, hazards, flooding, number of lanes, and combinations thereof.
15. A system for predicting road conditions and traffic volume, comprising:
a computer system including:
memory configured to store a graph of one or more road regions and program code, wherein each road region includes a plurality of road intersections and a plurality of road segments, and wherein the road intersections are represented as nodes in the graph and the road segments are represented as edged in the graph;
one or more processor devices operatively coupled to the memory for running the program code to:
translate the edges of the graph into nodes of a line graph;
embed the nodes from the graph into a node space and embed the nodes of the line graph into the node space;
to align the nodes from the line graph with the nodes from the graph using a learned mapping matrix, and optimize the alignment by utilizing a regularization term in a final loss function; and
output a set of node and edge representations that predicts the traffic flow for each of the road segments and road intersections of the graph based on the optimized alignment of the nodes from the line graph with the nodes from the graph.
16. The system as recited in claim 15 , wherein the one or more processor devices further run the program code to direct road improvements for the road segments and road intersections predicted to have the greatest traffic flows.
17. The system as recited in claim 15 , wherein the one or more processor devices further run program code to update the edge embeddings and node embedding by modeling meta-interactions between the nodes and edges of the graph.
18. The system as recited in claim 17 , wherein the meta-interactions include node-to-edge interactions, edge-to-node interactions, and node-to-node interactions.
19. The system as recited in claim 15 , wherein the graph is an attributed graphs represented as G=(V, E, X v , X E ), where V denotes the set of nodes, E denotes the set of edges, X v denotes the set of node attributes, and X E denotes the set of edge attributes.
20. The system as recited in claim 19 , wherein the edge attributes are only partially available.