DGCNN
DGCNN copied to clipboard
xueyunlong12589
Repetition code of the model for the paper "EEG Emotion Recognition Using Dynamical Graph Convolutional Neural Networks" in pytorch
Hi! I am using this model and I wonder if you knew the number of GCN used to create the DGCNN and the size of the first Fully Connected Layer....
Hello, I would like to know if the features are normalised or scaled or had any other type of preprocessing. Thanks in advance!
The paper proposed a novel way of learning the weight matrix, however it seems that spectral convolution cannot be applied to a directed graph since the eigenvectors corresponding to the...
原文中1\*1卷积层似乎是在Chebyshev项求和后用同一个卷积层进行的映射: $x\in (N, F)$，图卷积后 $x\in (N, F)$，然后在特征维度计算1\*1卷积 $x\in (N, K)$，K为卷积层输出维度。但代码里似乎对切比雪夫多项式的每一个部分都新建了一个1\*1卷积： https://github.com/xueyunlong12589/DGCNN/blob/a5392a6f091d0fc123a543d5bba26b7fa780bd62/model.py#L14 另外，Chebyshev多项式的系数似乎代码里没有体现？另外，切比雪夫多项式的计算公式好像和原文不一样？ https://github.com/xueyunlong12589/DGCNN/blob/a5392a6f091d0fc123a543d5bba26b7fa780bd62/utils.py#L31 原文公式是一个二阶表达式，代码里似乎是一阶的： $T_k(x) = 2x T_{k-1}(x) - T_{k-2}(x), k\ge 2$
Metadata
Owner
Metadata
Repetition code of the model for the paper "EEG Emotion Recognition Using Dynamical Graph Convolutional Neural Networks" in pytorch