Fig. 6
From: CacPred: a cascaded convolutional neural network for TF-DNA binding prediction
The framework of the CacPred model. The CacPred needs the forward sequences and reverse complementary sequences as inputs, which are encoded into a matrix, respectively. Each red and cadet blue wide-sided rectangle represents a convolution; Each red and cadet blue narrow-sided represents a transposed convolutional algorithm; each green wide-sided rectangle represents a concatenated convolution; σ is the sigmoid function