def main():
# static params
n_sample = 200
fit_mode = 'hebbian'
save_fig = False
for n_label, c in zip([2, 4], ['rb', 'gy']):
# dynamic params: noise_amount
noise_amounts = np.arange(0.0, 0.51, 0.05)
similarities = []
accuracies = []
for noise_amount in noise_amounts:
similarity, accuracy = hopfield_single_performance(n_sample,
n_label,
noise_amount,
fit_mode,
save_fig)
similarities.append(similarity)
accuracies.append(accuracy)
print('noise_amounts:', noise_amounts)
print('similarities:', similarities)
print('accuracies:', accuracies)
plt.plot(noise_amounts * 100, similarities,
label='similarities (n_label: {})'.format(n_label),
c=c[0], marker='o')
plt.plot(noise_amounts * 100, accuracies,
label='accuracies (n_label: {})'.format(n_label),
c=c[1], marker='o')
plt.xlabel('noise amount [%]')
plt.ylabel('performance')
plt.xticks(noise_amounts * 100)
plt.ylim(0, 1)
plt.legend(loc='lower left')
plt.savefig('noise_performance.png')
# static params
n_sample = 200
noise_amount = 0.05
fit_mode = 'hebbian'
save_fig = False
# dynamic params: n_label
n_labels = []
similarities = []
accuracies = []
for n_label in range(2, 7):
similarity, accuracy = hopfield_single_performance(n_sample,
n_label,
noise_amount,
fit_mode,
save_fig)
n_labels.append(n_label)
similarities.append(similarity)
accuracies.append(accuracy)
print('n_labels:', n_labels)
print('similarities:', similarities)
print('accuracies:', accuracies)
plt.plot(n_labels, similarities, label='similarities', c='r', marker='o')
plt.plot(n_labels, accuracies, label='accuracies', c='b', marker='o')
plt.xlabel('number of labels')
plt.ylabel('performance')
plt.xticks(n_labels)
