What is the prevalent problems in graphical neural networks is the problem of context propagation. During the propagation of content as is described by elite variable with the graph aggregation mechanism we often lose information or we suppress signal in favor of noise. Noise is often also propagated throughout neural networks when we try to do neighborhood aggregation within a note. These commonly known problems are usually addressed by sampling. We tend to ups and pull the signal pan down sample the noise and empirically determined whether this aggregation can be made more effective
Contextual Insight 