emb_val_x_s = embedder.fit_transform(val_x_s)
emb_val_x_t = embedder.fit_transform(val_x_t)
emb_test_x = embedder.fit_transform(test_x_t)
# =============================================================================
num_inputs = emb_x_s.shape[1]# input layer size
# =============================================================================
model_obj = my_models(num_inputs, dropout = dropout_pr)
model = model_obj.build_model()
model = model_obj.fit(emb_x_s, train_y_s, emb_val_x_s, val_y_s,
scale = NN_scaling)
model = model_obj.fit(emb_x_t, train_y_t, emb_val_x_t, val_y_t,
scale = NN_scaling)
error_fine_tuning, train_loss, val_loss, test_loss = \
model_obj.evaluate(emb_x_t,train_y_t, emb_val_x_t, val_y_t, emb_test_x,
test_y_t, scale = NN_scaling)
title = 'Naive learning + fine tuning'
print_stat(title, error_fine_tuning, train_loss, val_loss, test_loss)
# =============================================================================
from plotters import error_dist
error_dist(emb_x_s, train_y_s, emb_x_t, train_y_t, error_fine_tuning,
test_y_t, title = title)
plt.show()
from plotters import plot_cdf
plot_cdf(error_fine_tuning, 100)
plt.show()
ml_model_obj.evaluate(ml_x_t,train_y_t, ml_val_x_t, val_y_t, ml_test_x,
test_y_t, scale = NN_scaling)
if fine_tuning:
ml_model = ml_model_obj.fit(ml_x_t,
train_y_t,
ml_val_x_t,
val_y_t,
scale=NN_scaling)
error_metric_f, ml_train_loss_f, ml_val_loss_f, ml_test_loss_f = \
ml_model_obj.evaluate(ml_x_t,train_y_t, ml_val_x_t, val_y_t, ml_test_x,
test_y_t, scale = NN_scaling)
title = 'metric learning'
print_stat(title, error_metric, ml_train_loss, ml_val_loss, ml_test_loss)
if fine_tuning:
title = 'metric learning with fine tuning'
print_stat(title, error_metric_f, ml_train_loss_f, ml_val_loss_f,
ml_test_loss_f)
# =============================================================================
from plotters import error_dist
error_dist(ml_x_s,
train_y_s,
ml_x_t,
train_y_t,
error_metric,
test_y_t,
fine_tuning = True
%run -i naive_learning.py
fine_tuning = False
%run -i naive_learning.py
%run -i metric_learning_training.py
%run -i sample_selection_bias_by_unlabeled_tranining.py
%run -i transform_features.py
%run -i transformed_sample_selection.py
print("cat 1 no data from target location")
title = 'Naive learning'
print_stat(title, error_naive, nl_train_loss, nl_val_loss, nl_test_loss)
print("cat 2 additional labeled data set from target location")
title = 'Naive learning with fine-tuning'
print_stat(title, error_naive_f, nl_train_loss_f, nl_val_loss_f, nl_test_loss_f)
title = 'metric learning'
print_stat(title, error_metric, ml_train_loss, ml_val_loss, ml_test_loss)
print("cat 3 unlabeled data set from target location")
title = 'sample selection bias'
print_stat(title, error_sample_bias, ssbc_train_loss, ssbc_val_loss, ssbc_test_loss)
title = 'no labeled data - transformed sample selection bias'
print_stat(title, error_metric_plus_sample, t_ssbc_train_loss, t_ssbc_val_loss, t_ssbc_test_loss)
model_type='weighted',
dropout=dropout_pr)
model = w_model_obj.build_model()
model = w_model_obj.fit(emb_x_s,
train_y_s,
emb_val_x_s,
val_y_s,
val_w=training_val_weights,
scale=NN_scaling,
training_w=training_weights)
error_sample_bias, train_loss, val_loss, test_loss =\
w_model_obj.evaluate(emb_x_t, train_y_t, emb_val_x_t, val_y_t,
emb_test_x, test_y_t, scale = NN_scaling)
title = 'metric learning plus sample selection bias'
print_stat(title, error_sample_bias, train_loss, val_loss, test_loss)
# =============================================================================
def error_dist(X_s, X_t, error_test, weights=None, title=None):
from scipy.interpolate import griddata
rssi_interp_s = griddata(train_y_s, X_s, train_y_t, method='cubic')
rssi_diff = rssi_interp_s - X_t
rssi_diff = np.sum(rssi_diff, axis=1)
if weights is None:
fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(10, 5))
plot_scatter_colored(rssi_diff,
train_y_t,
title='measure difference',
ax=ax1)
plot_scatter_colored(error_test,
training_weights = np.ones(train_y_t.shape)
training_val_weights = np.ones(val_y_t.shape)
model = w_model_obj.fit(emb_x_t,
train_y_t,
emb_val_x_t,
val_y_t,
val_w=training_val_weights,
scale=NN_scaling,
training_w=training_weights)
error_metric_plus_sample_f, train_loss_f, val_loss_f, test_loss_f =\
w_model_obj.evaluate(emb_u_x_t, u_train_y_t, emb_u_val_x_t, u_val_y_t,
emb_test_x, test_y_t, scale = NN_scaling)
title = 'metric training plus sample selection bias'
print_stat(title, error_metric_plus_sample, train_loss, val_loss, test_loss)
if fine_tuning:
title = 'metric training plus sample selection bias'
print_stat(title, error_metric_plus_sample_f, train_loss_f, val_loss_f,
test_loss_f)
# =============================================================================
from plotters import error_dist
error_dist(emb_x_s,
train_y_s,
emb_u_x_t,
u_train_y_t,
error_metric_plus_sample,
test_y_t,
weights=coef_s,
del model_obj
model_obj = my_models(num_inputs, dropout=dropout_pr)
model = model_obj.build_model()
model = model_obj.fit(emb_x_t,
train_y_t,
emb_val_x_t,
val_y_t,
scale=NN_scaling)
error_normal, train_loss, val_loss, test_loss = \
model_obj.evaluate(emb_x_t,train_y_t, emb_val_x_t, val_y_t, emb_test_x,
test_y_t, scale = NN_scaling)
title = 'normal learning'
print_stat(title, error_normal, train_loss, val_loss, test_loss)
# =============================================================================
from plotters import error_dist
#error_dist(emb_x_s, train_y_s, emb_x_t, train_y_t, error_naive,
# test_y_t, title = title)
plt.show()
from plotters import plot_cdf
plot_cdf(error_normal, 100)
plt.show()
#
#del model
#del model_obj
scale = NN_scaling)
error_naive, nl_train_loss, nl_val_loss, nl_test_loss = \
model_obj.evaluate(emb_x_t,train_y_t, emb_val_x_t, val_y_t, emb_test_x,
test_y_t, scale = NN_scaling)
if fine_tuning:
model = model_obj.fit(emb_x_t, train_y_t, emb_val_x_t, val_y_t,
scale = NN_scaling)
error_naive_f, nl_train_loss_f, nl_val_loss_f, nl_test_loss_f = \
model_obj.evaluate(emb_x_t,train_y_t, emb_val_x_t, val_y_t, emb_test_x,
test_y_t, scale = NN_scaling)
title = 'Naive learning'
print_stat(title, error_naive, nl_train_loss, nl_val_loss, nl_test_loss)
if fine_tuning:
title = 'Naive learning with fine-tuning'
print_stat(title, error_naive_f, nl_train_loss_f, nl_val_loss_f, nl_test_loss_f)
# =============================================================================
from plotters import error_dist
error_dist(emb_x_s, train_y_s, emb_x_t, train_y_t, error_naive,
test_y_t, title = title)
plt.show()
from plotters import plot_cdf
plot_cdf(error_naive, 100)
plt.show()
del model
