GCNII

2000년 도입 · 논문 7편에서 사용

GCNII is an extension of a Graph Convolution Networks with two new techniques, initial residual and identify mapping, to tackle the problem of oversmoothing -- where stacking more layers and adding non-linearity tends to degrade performance. At each layer, initial residual constructs a skip connection from the input layer, while identity mapping adds an identity matrix to the weight matrix.

출처: Simple and Deep Graph Convolutional Networks

소개 논문: Simple and Deep Graph Convolutional Networks