A Robust Drug–Target Interaction Prediction Framework with Capsule Network and Transfer Learning

. 2023 Sep 14;24(18):14061. doi: 10.3390/ijms241814061

Algorithm 1: Dynamic Routing

Input:

{\hat{u}}_{j | i}

, r and l
Output: v_j

for all capsule i (primary capsule) in layer l and capsule j (class capsule) in layer (l + 1):
b_ij ⇐ 0
for r iterations do
for all capsules i in layer l: c_i ⇐ softmax(b_i)
for all capsules j in layer (l+1): s_j ⇐ $\sum_{i} c_{i j} {\hat{u}}_{j | i}$
for all capsules j in layer (l+1): v_j ⇐ squash(s_j)
for all capsule i in layer l and capsule j in layer (l+1):
b_ij ⇐ b_ij + ${\hat{u}}_{j | i} v_{j}$
end for
return v_j