Perceptron Logical OR ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

Question

I found another question with almost the same title, but it was about logical AND, and I have a problem with logical OR.

This is my code:

from numpy import array, random, dot
from random import choice
from pylab import ylim, plot

from matplotlib import pyplot as plt


def step_function(x): return 0 if x < 0 else 1


training_dataset = [

    (array([0, 1, 0.2345678]), 1),

    (array([1, 0, 1]), 1),

    (array([1, 1, 0]), 0),

    (array([1, 1, 0.5]), 1),

]

weights = random.rand(3)

error = []

learning_rate = 0.00000001

n = 100

for j in range(n):
    x, expected = choice(training_dataset)
    result = dot(weights, x)
    err = expected - step_function(result)
    error.append(err)
    weights = weights + learning_rate * err * x

for i in range(100):

    result = dot(i, weights)

    print("{}: {} -> {}".format(i, result, step_function(result)))

ylim([-1,1])

plot(error)

plt.show()

I was trying to make a perceptron. I don't know what the problem is, but ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all() happened when I changed the model evaluation loop from for i, _ in training_dataset: to for i in range(100):. I did that because it would show me the output only 4 times, and I wanted to see the program run a 100 times. Do you know why this is happening? How do I solve it?

Answer 1

The for-loop iterated over the training_dataset, not over a range. You need to supply the dot function with a training dataset.

from numpy import array, random, dot
from random import choice
from pylab import ylim, plot

from matplotlib import pyplot as plt


def step_function(x): return 0 if x < 0 else 1


training_dataset = [

    (array([0, 1, 0.2345678]), 1),

    (array([1, 0, 1]), 1),

    (array([1, 1, 0]), 0),

    (array([1, 1, 0.5]), 1),

]

weights = random.rand(3)

error = []

learning_rate = 0.1

n = 100

for j in range(n):
    x, expected = choice(training_dataset)
    result = dot(weights, x)
    err = expected - step_function(result)
    error.append(err)
    weights = weights + learning_rate * err * x

for i in range(100):

    k = random.randint(len(training_dataset))
    
    result = dot(training_dataset[k][0], weights)

    print("{}: {} -> {}".format(i, result, step_function(result)))

ylim([-1,1])

plot(error)

plt.show()