TheAlgorithms-Python

1
"""
2
Forward propagation explanation:
3
https://towardsdatascience.com/forward-propagation-in-neural-networks-simplified-math-and-code-version-bbcfef6f9250
4
"""
5

6
import math
7
import random
8

9

10
# Sigmoid
11
def sigmoid_function(value: float, deriv: bool = False) -> float:
12
    """Return the sigmoid function of a float.
13

14
    >>> sigmoid_function(3.5)
15
    0.9706877692486436
16
    >>> sigmoid_function(3.5, True)
17
    -8.75
18
    """
19
    if deriv:
20
        return value * (1 - value)
21
    return 1 / (1 + math.exp(-value))
22

23

24
# Initial Value
25
INITIAL_VALUE = 0.02
26

27

28
def forward_propagation(expected: int, number_propagations: int) -> float:
29
    """Return the value found after the forward propagation training.
30

31
    >>> res = forward_propagation(32, 450_000)  # Was 10_000_000
32
    >>> res > 31 and res < 33
33
    True
34

35
    >>> res = forward_propagation(32, 1000)
36
    >>> res > 31 and res < 33
37
    False
38
    """
39

40
    # Random weight
41
    weight = float(2 * (random.randint(1, 100)) - 1)
42

43
    for _ in range(number_propagations):
44
        # Forward propagation
45
        layer_1 = sigmoid_function(INITIAL_VALUE * weight)
46
        # How much did we miss?
47
        layer_1_error = (expected / 100) - layer_1
48
        # Error delta
49
        layer_1_delta = layer_1_error * sigmoid_function(layer_1, True)
50
        # Update weight
51
        weight += INITIAL_VALUE * layer_1_delta
52

53
    return layer_1 * 100
54

55

56
if __name__ == "__main__":
57
    import doctest
58

59
    doctest.testmod()
60

61
    expected = int(input("Expected value: "))
62
    number_propagations = int(input("Number of propagations: "))
63
    print(forward_propagation(expected, number_propagations))
64