Hello everyone!
I noticed that the source code (in pygod.detector) of this library for building benchmark models has a certain degree of commonality. For example, the benchmark models all inherit the DeepDetector class to build models and complete training reasoning, but the official documentation only gives the definitions of these modules, which brings me some difficulties.
I wonder if I can use this library to build my own model and complete training reasoning, that is, not just as an experimental script for the benchmark model.
Among many reconstruction-based graph anomaly detection papers, only a few papers have chosen open source. In these codes, the models seem to be built from scratch, and PyGOD is used to complete the experiments of the benchmark model.
I am a little confused in this regard. Can anyone answer my questions? Thank you!
Hello everyone!
I noticed that the source code (in
pygod.detector) of this library for building benchmark models has a certain degree of commonality. For example, the benchmark models all inherit theDeepDetectorclass to build models and complete training reasoning, but the official documentation only gives the definitions of these modules, which brings me some difficulties.I wonder if I can use this library to build my own model and complete training reasoning, that is, not just as an experimental script for the benchmark model.
Among many reconstruction-based graph anomaly detection papers, only a few papers have chosen open source. In these codes, the models seem to be built from scratch, and PyGOD is used to complete the experiments of the benchmark model.
I am a little confused in this regard. Can anyone answer my questions? Thank you!