Fig. 5
From: A gene regulatory network inference model based on pseudo-siamese network
Module effectiveness validation of PSGRN. This study aimed to verify the optimal performance of each module of PSGRN by comparing different modules. In order to evaluate the time feature learning module, we removed the GRU module and implemented either RNN or LSTM to replace it. Similarly, the spatial feature learning module was also substituted by either VGGNet or ResNet in order to assess its performance. Additionally, for part c, the loss function was replaced with the softmax function instead of the sigmoid function. These modifications were made to better understand the effectiveness of each module within PSGRN