"w"))
print("\n".join(map(str, history.history["loss"])),
file=open(
os.path.join(save_path, "train_losses{}.txt".format(i)),
"w"))
print("\n".join(map(str, history.history["val_acc"])),
file=open(
os.path.join(save_path, "val_accs{}.txt".format(i)),
"w"))
print("\n".join(map(str, history.history["val_loss"])),
file=open(
os.path.join(save_path, "val_losses{}.txt".format(i)),
"w"))
train_pred = model.predict_classes(X=Xs["train"],
batch_size=batch_size,
verbose=0)
train_perf = eval_pred(ys["train"], train_pred)
for met in ["acc", "prec", "rec", "f1"]:
av_train_perf[met] += float(
train_perf[met]) / args.ensemble_size
val_pred = model.predict_classes(X=Xs["val"],
batch_size=batch_size,
verbose=0)
val_perf = eval_pred(ys["val"], val_pred)
for met in ["acc", "prec", "rec", "f1"]:
av_val_perf[met] += float(val_perf[met]) / args.ensemble_size
final_train_perfs[params] = av_train_perf
final_val_perfs[params] = av_val_perf
history = model.fit(
X=np.concatenate((Xs["train"], Xs["val"])),
y=np.concatenate((ys["train"], ys["val"])),
batch_size=best_batch_size,
nb_epoch=best_epochs,
verbose=1,
shuffle=True,
show_accuracy=True,
)
model.save_weights(os.path.join(save_path, "weights{}.h5".format(i)), overwrite=True)
print("\n".join(map(str, history.history["acc"])), file=open(os.path.join(save_path, "train_accs{}.txt".format(i)), "w"))
print("\n".join(map(str, history.history["loss"])), file=open(os.path.join(save_path, "train_losses{}.txt".format(i)), "w"))
pred = model.predict_classes(X=Xs["test"], batch_size=batch_size, verbose=0)
preds[:, i] = pred[:, 0]
final_pred = mode(preds, axis=1).mode
test_perf = eval_pred(ys["test"], final_pred)
else:
print("Building model")
model = ShallowNet(Xs["train"].shape[1], best_dropout, best_dense_layers, best_dense_layer_units, args.weights)
model.compile(optimizer=Adam(lr=best_lr), loss="binary_crossentropy")
print("Model built")
test_perf = eval_model(model, best_batch_size, Xs["test"], ys["test"])
print("Test perf: {}".format(test_perf))
if args.train == "true":
shuffle=True,
show_accuracy=True,
)
model.save_weights(os.path.join(save_path, "weights{}.h5".format(i)),
overwrite=True)
print("\n".join(map(str, history.history["acc"])),
file=open(os.path.join(save_path, "train_accs{}.txt".format(i)),
"w"))
print("\n".join(map(str, history.history["loss"])),
file=open(
os.path.join(save_path, "train_losses{}.txt".format(i)),
"w"))
pred = model.predict_classes(X=Xs["test"],
batch_size=batch_size,
verbose=0)
preds[:, i] = pred[:, 0]
final_pred = mode(preds, axis=1).mode
test_perf = eval_pred(ys["test"], final_pred)
else:
print("Building model")
model = ShallowNet(Xs["train"].shape[1], best_dropout, best_dense_layers,
best_dense_layer_units, args.weights)
model.compile(optimizer=Adam(lr=best_lr), loss="binary_crossentropy")
print("Model built")
test_perf = eval_model(model, best_batch_size, Xs["test"], ys["test"])
print("Test perf: {}".format(test_perf))
y=ys["train"],
batch_size=batch_size,
nb_epoch=epochs,
verbose=1,
validation_data=(Xs["val"], ys["val"]),
shuffle=True,
show_accuracy=True,
callbacks=[LearningRateScheduler(lambda e: lr_schedule(epochs, lr, e))]
)
print("\n".join(map(str, history.history["acc"])), file=open(os.path.join(save_path, "train_accs{}.txt".format(i)), "w"))
print("\n".join(map(str, history.history["loss"])), file=open(os.path.join(save_path, "train_losses{}.txt".format(i)), "w"))
print("\n".join(map(str, history.history["val_acc"])), file=open(os.path.join(save_path, "val_accs{}.txt".format(i)), "w"))
print("\n".join(map(str, history.history["val_loss"])), file=open(os.path.join(save_path, "val_losses{}.txt".format(i)), "w"))
train_pred = model.predict_classes(X=Xs["train"], batch_size=batch_size, verbose=0)
train_perf = eval_pred(ys["train"], train_pred)
for met in ["acc", "prec", "rec", "f1"]:
av_train_perf[met] += float(train_perf[met]) / args.ensemble_size
val_pred = model.predict_classes(X=Xs["val"], batch_size=batch_size, verbose=0)
val_perf = eval_pred(ys["val"], val_pred)
for met in ["acc", "prec", "rec", "f1"]:
av_val_perf[met] += float(val_perf[met]) / args.ensemble_size
final_train_perfs[params] = av_train_perf
final_val_perfs[params] = av_val_perf
print("final train perf: acc {}, f1 {}; final val perf: acc {}, f1 {}".format(final_train_perfs[params]["acc"], final_train_perfs[params]["f1"], final_val_perfs[params]["acc"], final_val_perfs[params]["f1"]))
with open(os.path.join(base_save_dir, "checkpoints", "final_train_perfs_{}.pickle".format(turn)), "wb") as sf:
cPickle.dump(final_train_perfs, sf)
