def evaluator(x_train, y_train, x_val, y_val, x_test, y_test, experiment_name="", init="fixed", **kwargs):
# Define model
model = Sequential(loss=CrossEntropy())
model.add(Dense(nodes=10, input_dim=x_train.shape[0]))
model.add(Softmax())
# Fit model
model_save_path = "models/" + experiment_name + "/" + dict_to_string(kwargs) + "_" + init
best_model = model.fit(X=x_train, Y=y_train, X_val=x_val, Y_val=y_val,
save_path=model_save_path, **kwargs)
# Plot results
test_acc = best_model.get_classification_metrics(x_test, y_test)[0]
subtitle = "l2_reg: " + str(kwargs["l2_reg"]) + ", lr: " + str(kwargs["lr"]) +\
", weight_init:" + init + ", Test Acc: " + str(test_acc)
best_model.plot_training_progress(show=False,
save=True,
name="figures/" + experiment_name + "/" + dict_to_string(kwargs) + "_" + init,
subtitle=subtitle)
montage(W=np.array(best_model.layers[0].weights[:, :-1]),
title=subtitle,
path="figures/" + experiment_name + "/weights/" + dict_to_string(kwargs) + "_" + init)
# Minimizing value: validation accuracy
val_acc = best_model.get_classification_metrics(x_val, y_val)[0] # Get accuracy
result = {"value": val_acc, "model": best_model} # Save score and model
return result
def evaluator(x_train, y_train, x_val, y_val, **kwargs):
# Define model
model = Sequential(loss="cross_entropy")
model.add(
Dense(nodes=10, input_dim=x_train.shape[0], weight_initialization="fixed"))
model.add(Activation("softmax"))
# Fit model
model.fit(X=x_train, Y=y_train, X_val=x_val, Y_val=y_val, **kwargs)
model.plot_training_progress(show=False, save=True, name="figures/" + dict_to_string(kwargs))
# model.save("models/" + dict_to_string(kwargs))
# Minimizing value:
value = model.get_classification_metrics(x_val, y_val)[0] # Get accuracy
result = {"value": value, "model": model} # Save score and model
return result
def evaluator(l2_reg):
# Define model
model = Sequential(loss=CrossEntropy(), metric=Accuracy())
model.add(Dense(nodes=800, input_dim=x_train.shape[0]))
model.add(Relu())
model.add(Dense(nodes=10, input_dim=800))
model.add(Softmax())
ns = 800
# Define callbacks
mt = MetricTracker() # Stores training evolution info
lrs = LearningRateScheduler(evolution="cyclic",
lr_min=1e-3,
lr_max=1e-1,
ns=ns) # Modifies lr while training
callbacks = [mt, lrs]
# Fit model
iterations = 4 * ns
model.fit(X=x_train,
Y=y_train,
X_val=x_val,
Y_val=y_val,
batch_size=100,
iterations=iterations,
l2_reg=l2_reg,
shuffle_minibatch=True,
callbacks=callbacks)
model.save("models/yes_dropout_test")
# Test model
val_acc = model.get_metric_loss(x_val, y_val)[0]
test_acc = model.get_metric_loss(x_test, y_test)[0]
subtitle = "L2 param: " + str(l2_reg) + ", Test acc: " + str(test_acc)
mt.plot_training_progress(show=True,
save=True,
name="figures/l2reg_optimization/" + str(l2_reg),
subtitle=subtitle)
print("Val accuracy:", val_acc)
print("Test accuracy:", test_acc)
return val_acc
if __name__ == "__main__":
x_train, y_train, x_val, y_val, x_test, y_test = read_cifar_10(n_train=3, n_val=5, n_test=2)
# x_train, y_train, x_val, y_val, x_test, y_test = read_mnist(n_train=2, n_val=5, n_test=2)
# x_train, y_train, x_val, y_val, x_test, y_test = read_names(n_train=500)
class_sum = np.sum(y_train, axis=1)*y_train.shape[0]
class_count = np.reciprocal(class_sum, where=abs(class_sum) > 0)
print(class_count)
print(type(x_train[0, 0, 0]))
# Define model
model = Sequential(loss=CrossEntropy(), metric=Accuracy())
model.add(Conv2D(num_filters=2, kernel_shape=(4, 4), stride=3, dilation_rate=2, input_shape=x_train.shape[0:-1]))
model.add(Relu())
model.add(MaxPool2D((2, 2), stride=3))
model.add(Flatten())
model.add(Dense(nodes=y_train.shape[0]))
model.add(Relu())
model.add(Softmax())
print(np.min(np.abs(model.layers[0].filters)))
reg = 0.0
# Fit model
anal_time = time.time()
model.fit(X=x_train, Y=y_train, X_val=x_val, Y_val=y_val,
batch_size=200, epochs=1, lr=0, momentum=0, l2_reg=reg)
np.abs(analytical_grad_weight-numerical_grad_w),
np.multiply(denom, (denom > _EPS)) + np.multiply(_EPS*np.ones(denom.shape), (denom <= _EPS)))
np.set_printoptions(suppress=True)
print(np.round(a*100,decimals=2))
av_error = np.average(a)
max_error = np.max(a)
print("Averaged Element-Wise Relative Error:", av_error*100, "%")
print("Max Element-Wise Relative Error:", max_error*100, "%")
# np.set_printoptions(suppress=False)
if __name__ == "__main__":
# Define model
v_rnn = VanillaRNN(state_size=state_size, input_size=K, output_size=K)
model = Sequential(loss=CrossEntropy(class_count=None), metric=Accuracy())
model.add(v_rnn)
model.layers[0].reset_state(copy.deepcopy(state))
# print(model.layers[0].c)
# Fit model
l2_reg = 0.0
model.fit(X=encoded_data, epochs=1, lr = 2e-2, momentum=0.95, l2_reg=l2_reg,
batcher=RnnBatcher(seq_length), callbacks=[])
print(model.layers[0].dl_dc)
anal = copy.deepcopy(model.layers[0].dl_dc)
model.layers[0].reset_state(copy.deepcopy(state))
x = np.array((char_to_ind['o'], char_to_ind['l'], char_to_ind['i'], ))
# x_test, y_test = getXY(LoadBatch("test_batch"))
# Preprocessing
mean_x = np.mean(x_train)
std_x = np.std(x_train)
x_train = (x_train - mean_x)/std_x
# x_val = (x_val - mean_x)/std_x
# x_test = (x_test - mean_x)/std_x
x = x_train[:, 0:5]
y = y_train[:, 0:5]
reg = 0.1
# Define model
model = Sequential(loss="categorical_hinge")
model.add(Dense(nodes=10, input_dim=x.shape[0], weight_initialization="fixed"))
anal_time = time.time()
model.fit(x, y,
batch_size=10000, epochs=1, lr=0, # 0 lr will not change weights
momentum=0, l2_reg=reg)
analytical_grad = model.layers[0].gradient
anal_time = anal_time - time.time()
# Get Numerical gradient
num_time = time.time()
numerical_grad = ComputeGradsNum(x, y, model, l2_reg=reg, h=0.001)
print(numerical_grad.shape)
num_time = num_time - time.time()
_EPS = 0.0000001
mt = MetricTracker() # Stores training evolution info (losses and metrics)
# lrs = LearningRateScheduler(evolution="linear", lr_min=1e-3, lr_max=9e-1)
# lrs = LearningRateScheduler(evolution="constant", lr_min=1e-3, lr_max=9e-1)
# callbacks = [mt, lrs]
callbacks = [mt]
# Define hyperparams
d = x_train.shape[0]
n1 = 40 # Filters of first Conv2D
k1 = 6 # First kernel y size
n2 = 20 # Filters of second Conv2D
k2 = 4 # Second kernel y size
# Define model
model = Sequential(loss=CrossEntropy(class_count=None), metric=Accuracy())
model.add(
Conv2D(num_filters=n1,
kernel_shape=(d, k1),
input_shape=x_train.shape[:-1]))
model.add(Relu())
model.add(Conv2D(num_filters=n2, kernel_shape=(1, k2)))
model.add(Relu())
model.add(Flatten())
model.add(Dense(nodes=y_train.shape[0]))
model.add(Softmax())
# Fit model
model.fit(X=x_train,
Y=y_train,
X_val=x_val,
Y_val=y_val,
batch_size=100,
epochs=500,
lr=1e-3,
def evaluator(x_train, y_train, x_val, y_val, experiment_name="", **kwargs):
print(kwargs)
# Saving directories
figure_file = "figures/" + experiment_name + "/" + dict_to_string(kwargs)
model_file = "models/" + experiment_name + "/" + dict_to_string(kwargs)
mt = MetricTracker() # Stores training evolution info (losses and metrics)
# Define model
d = x_train.shape[0]
n1 = kwargs["n1"] # Filters of first Conv2D
k1 = kwargs["k1"] # First kernel y size
n2 = kwargs["n2"] # Filters of second Conv2D
k2 = kwargs["k2"] # Second kernel y size
batch_size = kwargs["batch_size"]
try:
# Define model
model = Sequential(loss=CrossEntropy(class_count=None),
metric=Accuracy())
model.add(
Conv2D(num_filters=n1,
kernel_shape=(d, k1),
input_shape=x_train.shape[:-1]))
model.add(Relu())
model.add(Conv2D(num_filters=n2, kernel_shape=(1, k2)))
model.add(Relu())
model.add(Flatten())
model.add(Dense(nodes=y_train.shape[0]))
model.add(Softmax())
# Fit model
model.fit(X=x_train,
Y=y_train,
X_val=x_val,
Y_val=y_val,
batch_size=batch_size,
epochs=1000,
lr=1e-2,
momentum=0.8,
l2_reg=0.001,
compensate=True,
callbacks=[mt])
except Exception as e:
print(e)
return -1 # If configuration impossible
model.save(model_file)
# Write results
n1 = str(n1)
n2 = str(n2)
k1 = str(k1)
k2 = str(k2)
batch_size = str(batch_size)
subtitle = "n1:" + n1 + ", n2:" + n2 + ", k1:" + k1 + ", k2:" + k1 +\
", batch_size:" + batch_size
mt.plot_training_progress(show=False,
save=True,
name=figure_file,
subtitle=subtitle)
# Maximizing value: validation accuracy
return model.val_metric
if __name__ == "__main__":
# Load data
x_train, y_train = LoadXY("data_batch_1")
x_val, y_val = LoadXY("data_batch_2")
x_test, y_test = LoadXY("test_batch")
# Preprocessing
mean_x = np.mean(x_train)
std_x = np.std(x_train)
x_train = (x_train - mean_x) / std_x
x_val = (x_val - mean_x) / std_x
x_test = (x_test - mean_x) / std_x
# Define model
model = Sequential(loss=CrossEntropy(), metric=Accuracy())
model.add(Dense(nodes=800, input_dim=x_train.shape[0]))
model.add(Relu())
model.add(Dropout(ones_ratio=0.50))
model.add(Dense(nodes=10, input_dim=800))
model.add(Softmax())
ns = 500
# Define callbacks
mt = MetricTracker() # Stores training evolution info
# bms = BestModelSaver(save_dir=None) # Saves model with highest val_metric
lrs = LearningRateScheduler(evolution="cyclic",
lr_min=1e-3,
lr_max=1e-1,
ns=ns) # Modifies lr while training
# callbacks = [mt, bms, lrs]
def evaluator(x_train,
y_train,
x_val,
y_val,
x_test,
y_test,
experiment_name="",
**kwargs):
# Saving directories
figure_file = "figures/" + experiment_name + "/" + dict_to_string(kwargs)
model_file = "models/" + experiment_name + "/" + dict_to_string(kwargs)
# Define model
model = Sequential(loss=CrossEntropy(), metric=Accuracy())
model.add(Dense(nodes=50, input_dim=x_train.shape[0]))
model.add(Relu())
model.add(Dense(nodes=10, input_dim=50))
model.add(Softmax())
# Pick metaparams
batch_size = 100
ns = 2 * np.floor(x_train.shape[1] / batch_size)
iterations = 4 * ns # 2 cycles
# Define callbacks
mt = MetricTracker() # Stores training evolution info
# bms = BestModelSaver(save_dir=None)
lrs = LearningRateScheduler(evolution="cyclic",
lr_min=1e-5,
lr_max=1e-1,
ns=ns)
# callbacks = [mt, bms, lrs]
callbacks = [mt, lrs]
# Adjust logarithmic
kwargs["l2_reg"] = 10**kwargs["l2_reg"]
# Fit model
model.fit(X=x_train,
Y=y_train,
X_val=x_val,
Y_val=y_val,
batch_size=batch_size,
epochs=None,
iterations=iterations,
**kwargs,
callbacks=callbacks)
# Write results
# best_model = bms.get_best_model()
test_acc = model.get_metric_loss(x_test, y_test)[0]
subtitle = "l2_reg: " + str(
kwargs["l2_reg"]) + ", Test Acc: " + str(test_acc)
mt.plot_training_progress(show=False,
save=True,
name=figure_file,
subtitle=subtitle)
# Maximizing value: validation accuracy
# val_metric = bms.best_metric
val_metric = model.get_metric_loss(x_val, y_val)[0]
return val_metric
# Load data
x_train, y_train = getXY(LoadBatch("data_batch_1"))
x_val, y_val = getXY(LoadBatch("data_batch_2"))
x_test, y_test = getXY(LoadBatch("test_batch"))
# Preprocessing
mean_x = np.mean(x_train)
std_x = np.std(x_train)
x_train = (x_train - mean_x) / std_x
x_val = (x_val - mean_x) / std_x
x_test = (x_test - mean_x) / std_x
# Define SVM multi-class model
model = Sequential(loss="categorical_hinge")
model.add(Dense(nodes=10, input_dim=x_train.shape[0]))
model.add(Activation("softmax"))
# Fit model
model.fit(X=x_train,
Y=y_train,
X_val=x_val,
Y_val=y_val,
batch_size=100,
epochs=100,
lr=0.0001,
momentum=0.1,
l2_reg=0.1)
model.plot_training_progress()
# Test model
(denom > _EPS)) + np.multiply(_EPS * np.ones(denom.shape),
(denom <= _EPS)))
np.set_printoptions(suppress=True)
print(np.round(a * 100, decimals=2))
av_error = np.average(a)
max_error = np.max(a)
print("Averaged Element-Wise Relative Error:", av_error * 100, "%")
print("Max Element-Wise Relative Error:", max_error * 100, "%")
# np.set_printoptions(suppress=False)
if __name__ == "__main__":
# Define model
v_rnn = VanillaRNN(state_size=state_size, input_size=K, output_size=K)
model = Sequential(loss=CrossEntropy(class_count=None), metric=Accuracy())
model.add(v_rnn)
model.add(Softmax())
# state = np.array(np.random.normal(0.1, 1./100.,
# (v_rnn.state_size,1)))
model.layers[0].reset_state(copy.deepcopy(state))
print(model.layers[0].W)
# Fit model
l2_reg = 0.0
t1 = time.time()
model.fit(X=encoded_data,
epochs=1,
lr=0,
l2_reg=l2_reg,
batcher=RnnBatcher(seq_length),
# for i in range(200):
# cv2.imshow("image", x_train[..., i])
# cv2.waitKey()
# Define callbacks
mt = MetricTracker(file_name="cifar_test_3") # Stores training evolution info (losses and metrics)
# bms = BestModelSaver("models/best_cifar") # Stores training evolution info (losses and metrics)
lrs = LearningRateScheduler(evolution="cyclic", lr_min=1e-3, lr_max=0.2, ns=500)
# lrs = LearningRateScheduler(evolution="constant", lr_min=1e-3, lr_max=9e-1)
# callbacks = [mt, lrs]
callbacks = [mt, lrs]
# Define architecture (copied from https://appliedmachinelearning.blog/2018/03/24/achieving-90-accuracy-in-object-recognition-task-on-cifar-10-dataset-with-keras-convolutional-neural-networks/)
model = Sequential(loss=CrossEntropy(), metric=Accuracy())
model.add(Conv2D(num_filters=32, kernel_shape=(3, 3), stride=2, input_shape=(32, 32, 3)))
model.add(Relu())
model.add(Conv2D(num_filters=64, kernel_shape=(3, 3)))
model.add(Relu())
model.add(MaxPool2D(kernel_shape=(2, 2), stride=2))
model.add(Conv2D(num_filters=128, kernel_shape=(2, 2)))
model.add(Relu())
model.add(MaxPool2D(kernel_shape=(2, 2)))
model.add(Flatten())
model.add(Dense(nodes=200))
model.add(Relu())
model.add(Dense(nodes=10))
model.add(Softmax())
# for filt in model.layers[0].filters:
if __name__ == "__main__":
# Load data
x_train, y_train = LoadXY("data_batch_1")
x_val, y_val = LoadXY("data_batch_2")
x_test, y_test = LoadXY("test_batch")
# Preprocessing
mean_x = np.mean(x_train)
std_x = np.std(x_train)
x_train = (x_train - mean_x) / std_x
x_val = (x_val - mean_x) / std_x
x_test = (x_test - mean_x) / std_x
# Define model
model = Sequential(loss=CrossEntropy(), metric=Accuracy())
model.add(Dense(nodes=50, input_dim=x_train.shape[0]))
model.add(Relu())
model.add(Dense(nodes=10, input_dim=50))
model.add(Softmax())
ns = 800
# Define callbacks
mt = MetricTracker() # Stores training evolution info
lrs = LearningRateScheduler(evolution="cyclic",
lr_min=1e-7,
lr_max=1e-2,
ns=ns) # Modifies lr while training
callbacks = [mt, lrs]
# Fit model
if __name__ == "__main__":
# Load data
x_train, y_train, x_val, y_val, x_test, y_test = read_mnist(n_train=200,
n_val=200,
n_test=2)
# Define callbacks
mt = MetricTracker() # Stores training evolution info (losses and metrics)
# lrs = LearningRateScheduler(evolution="linear", lr_min=1e-3, lr_max=9e-1)
# lrs = LearningRateScheduler(evolution="constant", lr_min=1e-3, lr_max=9e-1)
# callbacks = [mt, lrs]
callbacks = [mt]
# Define model
model = Sequential(loss=CrossEntropy(), metric=Accuracy())
model.add(
Conv2D(num_filters=64, kernel_shape=(4, 4), input_shape=(28, 28, 1)))
model.add(Relu())
model.add(MaxPool2D(kernel_shape=(2, 2)))
# model.add(Conv2D(num_filters=32, kernel_shape=(3, 3)))
# model.add(Relu())
model.add(Flatten())
# model.add(Flatten(input_shape=(28, 28, 1)))
model.add(Dense(nodes=400))
model.add(Relu())
model.add(Dense(nodes=10))
model.add(Softmax())
# for filt in model.layers[0].filters:
# print(filt)
# y_pred_prob = model.predict(x_train)
# print(y_pred_prob)
