xianggebenben / ivgd

ivgd's Issues

diffusion processing

This algorithm is based on the fact that the source detection is found in the case of the scattered model. So when comparing the experiment, is other models also using graph residual networks to disseminate simulation? But I see that the pre -training diagram diffusion model in your code uses MLP for training, so I don't understand this place very well.

Inquiry Regarding Data Generation Details

I am particularly interested in the technical details of the data generation.

What concerns me is that according to my research, most current papers regarding source localization do not contain the settings of the simulation (e.g. the choice of infection probability/recovery probability in SIR; are they heterogeneous, or all the edges share the same prob;) for the cascade generation in datasets used as basement in training and testing.

I am particularly interested in the technical details of the data simulation. Could you kindly share the relevant codes or information about the choice of specific parameters that may help me understand the process of data generation?