Skip to main content
. 2022 Jan 3;15:1. doi: 10.1186/s13040-021-00285-4

Fig. 4.

Fig. 4

Encoding information in the DNN. a. The proposed network is based on a feedforward neural network. b. The integration of biological knowledge is implemented into the first hidden layer, i.e. each neuron/node corresponds to one biological unit. c. The learned representation or encoding (the activation of the last hidden layer) can be used as the input of the TSNE algorithm to produce a 2D visualization of the data. Finally, the supervised performance of the network can be evaluated with classical ML metrics