Skip to main content

Table 3 Model structure and parameters for drug sequence information extraction

From: SSF-DDI: a deep learning method utilizing drug sequence and substructure features for drug–drug interaction prediction

Layer name

Output size

Parameters

Embedding Layer

[batch_size, 100,64 ]

num_embeddings 65, embedding dim 64

Conv1

[batch_size,40,97]

In_channels 64, out_channels 40, kernel 4, stride 1

Conv2

[batch_size,80,92]

In_channels 40, out_channels 80, kernel 6, stride 1

Conv3

[batch_size,160,85]

In_channels 80, out_channels 160, kernel 8, stride 1

MixAttention Layer

[batch_size,160,85]

In_features 160, out_features 160

Maxpooling Layer

[batch_size,160]

Kernel_size 85, stride 85