naveenkumarmarri / graph-convolutions Goto Github PK

all mySQL files are setup in C:\ProgramData\MySQL\MySQL Server 8.0\Data\msem_db

from msql console, use execute the following command:

D:\projects\graph_analysis\DataloadScript.sql

You can modify DataloadScript to add more columns such as contrast, brightness, etc. You also need to add index to these new added parameters.

// graph based analysis tools use klayout to open and view write_buffer.gds, which contains the circuit layout info

python read_from_gds.py // internally has the path to write_buffer.gds this decode info from gds file and write to a file: data/gds_polygon_and_label_try.txt you need to copy the above file to C:\ProgramData\MySQL\MySQL Server 8.0\Data\msem_db

from mysql console, run extract_relationships.sql which will load in data/gds_polygon_and_label_try.txt and compute all spatial overlap between polygons outputs to C:\ProgramData\MySQL\MySQL Server 8.0\Data\msem_db are:

node_features_length_reordered.txt node_labels_final_reordered.txt relationship_final_mysql_nodes_reordered.txt

all GCN model and data are saved in the folder: graph_analysis/GCN

ind.gds.tx and ind.gds.ty are the test data for GCN ind.gds.x and ind.gds.y are the training data for GCN to generate these, run the following command to install the GCN package

python setup.py To invoke the data processing pipeline for the algorithm python dataparse_gcn.py

to train and test, run GCN/train.py The model is saved in models.python

2nd approach is to use deepwalk, which takes relationship_final_mysql_nodes_reordered.txt

python deepwalk_fullyconnected.py

graph_clustering_gds is a clustering method using autoencoder-decoder approach input is the similarity matrix output