Pooling Operation

For the data computation, pooling could reduce the overall loading for afterward works efficiently. The following examples show the ordinary operations of pooling.

For the upper case, max pooling is implemented.
For the lower case, average pooling is implemented.



In [1]:

    
%matplotlib inline
from lab006 import *
import numpy as np
import matplotlib.pyplot as plt
# Question 1
target = list(mnist_read('train-images-idx3-ubyte', 'train-labels-idx1-ubyte'))

plt.figure()
for i in range(5):
    label, img = target[i]
    # Max Pooling
    maxPoolResult = max_pooling(img)
    maxPoolResult.astype(np.uint8)
    plt.subplot(2, 5, i+1)
    mnist_show(maxPoolResult)

    # Average Pooling
    avgPoolResult = avg_pooling(img)
    avgPoolResult.astype(np.uint8)
    plt.subplot(2, 5, i+6)
    mnist_show(avgPoolResult)

plt.show()