4.3 Limitations

Four Limitations

1. Computationally inefficient to update the hidden states of nodes iteratively to get the fixed point.
2. uses the same parameters in the iteration while most opoular neural networks use different parameters in different layers, which serves as a hierarchical feature extraction method
3. some informative edges(Knowledge Graph) cannot be effectively modeled in GNN.
4. if T is pretty large, it is unsuitable to use the fixed points if we focus on the representation of nodes instead of graphs because the distribution of representation in the fixed point will be much more smooth in value and less informative for distinguishihng each node.!