Perceptron

perceptron.py

def activation(output):
    """
    Activation function

    :param output: Perceptron algorithm result
    :return: 1 if output >= 0 and -1 if output < 0
    """
    if output >= 0:
        res = 1
    else:
        res = -1
    return res


def perceptron(xs, ws):
    """
    Perceptron Algorithm

    :param xs: values for training
    :param ws: weights for each value
    :return: a possible class
    """
    output = 0
    for x, w in zip(xs, ws):
        output += x * w
    return activation(output)


def calculate_weights(inputs, weights, n, e):
    """
    Calculate weights

    :param inputs: current Entries
    :param weights: current weights
    :param n: learning rate
    :param e: error
    :return: recalculated weights
    """
    for i, w in enumerate(weights):
        weights[i] = w + n * e * inputs[i]


def training(data_set, n, bias=1):
    """
    Trains the perceptron algorithm

    :param data_set: Dataset for training
    :param n: learning rate
    :param bias: A threshold
    :return: Prints errors and hits
    """
    count = 0
    weights = None

    while len(data_set) > count:
        # Separate the entries of the expected classes
        *inputs, ce = data_set[count]
        # Add bias to entries
        inputs = inputs + [bias]
        # Generates weights for each entry if it does not exist
        if not weights:
            weights = [0.0 for _ in inputs]
        # Predict the class
        cp = perceptron(inputs, weights)
        # Calculate the error
        e = ce - cp
        # Check if the RNA is correct
        if e == 0:
            # Prints that is correct
            print(f'Input {count + 1} is correct')
            # Pass to next entry
            count += 1
        else:
            # Recalculate weights
            calculate_weights(inputs, weights, n, e)
            # Prints that is incorrect
            print(f'Input {count + 1} is incorrect. Adjusting the weights...')
            # Back to first entry
            count = 0


if __name__ == '__main__':
    # x1 | x2 | class
    examples = (
        (1.0, 1.0, 1),
        (9.4, 6.4, -1),
        (2.5, 2.1, 1),
        (8.0, 7.7, -1),
        (0.5, 2.2, 1),
        (7.9, 8.4, -1),
        (7.0, 7.0, -1),
        (2.8, 0.8, 1),
        (1.2, 3.0, 1),
        (7.8, 6.1, -1),
    )
    training(examples, n=0.2)