Dynamic Learning Rate in Deep CNN Model for Metastasis Detection and Classification of Histopathology Images

Algorithm 4

Pseudocode for the proposed CNN model.

Input: input image

Output: binary classification

Data: image, epochs, batch size

Result: classification with prediction

1 initialization;

2 whiledo

3 extract features;

4 foreach epochdo

5  train ;

6  foreach minibatchdo

   / feature extraction: layer-1

7   extract low-level features;

8   perform dimensionality reduction (max pooling);

   / feature extraction: layer-2

9   extract high-level features;

10   perform dimensionality reduction (max pooling);

   / feature extraction: layer-3

11   extract high-level features;

12   perform dimensionality reduction (max pooling);

   / flatten layer

13   feature vector arranged as a one-dimensional array;

   / classification layer

14   two fully connected layers performs classification;

15  end

16  calculate average loss over each epoch in a minibatch;

17  backpropagation applied to every iteration;

18 end

19 ;

20 end