Algorithm 2: Pruning Neurons.

1: Input: The model weights W, the prune rate α, the number of prune iterations iter

2: Get $m$ which is the number of elements in W

3: Get the number of W that need pruning $n=\alpha \ast m$

4: Calculate the importance of neurons, get the neuron importance score matrix $NIS$

5: Sorting $NIS$ from large to small

6: Get the prune threshold of neuron importance score $thr=NIS\left[n\right]$

7: While i < m do

8:  if $NIS\left[i\right]$ < thr then

9:  W$\left[i\right]$ = 0

10: end

11: $i++$

12: end

13: While k < iter do

14:  Retrain the model to update parameters $W$ according to Algorithm 1

15:  Execute step 2–12

16:  $k++$

17: end