The code of graph adnomal detection project
Pytorch
Tensorflow (Tensorboard)
main.py contains parameters needed by the use. It is a good start if you want to get yourself familiar with code. DGAD.py contains trainning and testing process. The 3D graph convolution is defined in d3_graph_conv.py and net.py contains the whole network class. trans_graph.py is used to transform the .csv dataset to .npz gaph format.
The path to the input dataset can be changed in DGAD.py.
Lis_GAD.py and Lis_net.py contains the method in previous work.
Dataset, related paper and our slide can be found on https://drive.google.com/drive/folders/1zXSLLmbTnJhLSYeSZ41N8ZNcGfH2VmLn?usp=sharing
Technical Reports is on https://www.overleaf.com/read/nddvkswmdrjr
Add code comments and clean code.
Improve detection result with effectives and efficience: better covolution operation, integrate cluster gcn, new anmolay measure (cluster)