for m in metrics:
accumulated_metrics[m] = []
name = f"MLP{i:05}_d-{dataset_idx}_b-{batch_size}_h-{n_hidden_nodes}_eta-{eta}".replace(".", ",")
i += 1
print(name)
best = {"best_acc": 0, "best_loss": np.inf}
for _ in range(loops):
print(".", end="")
net = MLP(train[0], train[1], n_hidden_nodes, momentum=0)
train_losses, valid_losses, train_accuracies, valid_accuracies, pocket_epoch = net.train(
train[0], train[1], valid[0], valid[1], eta, epochs,
early_stop_count=early_stop, shuffle=shuffle, batch_size=batch_size
)
cm_train = net.confmat(train[0], train[1])
cm_valid = net.confmat(valid[0], valid[1])
print("cm_train", cm_train)
print("cm_valid", cm_valid)
print("pocket epoch", pocket_epoch)
train_loss, valid_loss = train_losses[pocket_epoch], valid_losses[pocket_epoch]
train_acc, train_sens, train_spec, _, _ = two_class_conf_mat_metrics(
net.confmat(train[0], train[1]))
valid_acc, valid_sens, valid_spec, _, _ = two_class_conf_mat_metrics(
net.confmat(valid[0], valid[1]))
if debug_best_and_plot:
if valid_acc > best["best_acc"] or (
valid_acc == best["best_acc"] and valid_loss < best["best_loss"]):
print("BEST!")
import numpy as np
from model.mlp import MLP
if __name__ == '__main__':
np.random.seed(72)
anddata = np.array([[0, 0, 0], [0, 1, 0], [1, 0, 0], [1, 1, 1]])
xordata = np.array([[0, 0, 0], [0, 1, 1], [1, 0, 1], [1, 1, 0]])
p = MLP(anddata[:, 0:2], anddata[:, 2:3], 2)
p.train_for_niterations(anddata[:, 0:2], anddata[:, 2:3], 0.25, 1001)
p.confmat(anddata[:, 0:2], anddata[:, 2:3])
q = MLP(xordata[:, 0:2], xordata[:, 2:3], 2)
q.train_for_niterations(xordata[:, 0:2], xordata[:, 2:3], 0.25, 5001)
q.confmat(xordata[:, 0:2], xordata[:, 2:3])
valid = standardize_dataset(valid, mean, std)
test = standardize_dataset(test, mean, std)
# for i in [1, 2, 5, 10, 20, 100]:
for i in [20]:
print(f"x-----> {i} hidden nodes <-----x")
net = MLP(train[0], number_to_one_hoot(train[1]), i, outtype="softmax")
# net.train(train[0], number_to_one_hoot(train[1]), 0.1, 1000)
net.earlystopping_primitive(train[0],
number_to_one_hoot(train[1]),
valid[0],
number_to_one_hoot(valid[1]),
eta=0.1,
niteration=100,
early_stop_count=2)
net.confmat(train[0], number_to_one_hoot(train[1]))
net.confmat(test[0], number_to_one_hoot(test[1]))
print()
print()
"""
Confusion matrix:
[[4827 0 15 6 13 16 23 5 15 21]
[ 1 5541 22 12 7 8 9 12 37 8]
[ 10 25 4700 60 15 5 13 38 24 18]
[ 8 24 41 4780 2 81 2 18 51 31]
[ 6 6 40 2 4659 25 13 34 12 71]
[ 23 19 18 90 4 4260 46 7 40 27]
[ 19 3 26 10 46 31 4826 5 19 6]
[ 5 13 39 44 8 8 2 4965 8 92]
[ 30 33 51 68 10 52 17 11 4614 42]
[ 3 14 16 29 95 20 0 80 22 4672]]
print(np.abs(inputs).max(0))
# Targets to one hoot
targets = number_to_one_hoot(targets, 3)
train = (inputs[::2], targets[::2])
valid = (inputs[1::4], targets[1::4])
test = (inputs[3::4], targets[3::4])
accs = []
for i in range(10):
net = MLP(train[0], train[1], 10)
net.earlystopping_primitive(train[0], train[1], valid[0], valid[1], 0.1, 10, 2)
# net.train(train[0], train[1], 0.1, 100000)
print("Train")
net.confmat(train[0], train[1])
print("Valid")
net.confmat(valid[0], valid[1])
print("Test")
acc = conf_mat_acc(net.confmat(test[0], test[1]))
accs.append(acc)
print(f"Mean accuracy: {np.mean(np.array(accs)) * 100:2.4f}")
"""
5. Number of Instances: 150 (50 in each of three classes)
6. Number of Attributes: 4 numeric, predictive attributes and the class
7. Attribute Information:
1. sepal length in cm
2. sepal width in cm
3. petal length in cm
