def run_nn_dmi(args):
set_global_seeds(args['seed'])
dataset = DataLoader(args['dataset'], args)
X_train, X_test, X_val, y_train, y_test, y_val = dataset.prepare_train_test_val(
args)
mlp = MLP(
feature_dim=X_train.shape[-1],
hidsizes=args['hidsize'],
dropout=args['dropout'],
outputs=2,
)
classifier = DMIClassifier(
model=mlp,
learning_rate=args['lr'],
)
results = classifier.fit(
X_train,
y_train,
X_test,
y_test,
batchsize=args['batchsize'],
episodes=args['episodes'],
logger=logger if args['seeds'] == 1 else None,
)
return results
def run_c_svm(args):
set_global_seeds(args['seed'])
dataset = DataLoader(args['dataset'], args)
X_train, X_test, X_val, y_train, y_test, y_val = dataset.prepare_train_test_val(
args)
model = SVC(gamma='auto', class_weight={0: 1., 1: args['C1']})
model.fit(X_train, y_train)
return model.score(X_test, y_test)
def run_pam(args):
set_global_seeds(args['seed'])
dataset = DataLoader(args['dataset'])
X_train, X_test, X_val, y_train, y_test, y_val = dataset.prepare_train_test_val(
args)
model = Perceptron(feature_dim=X_train.shape[-1], margin=args['margin'])
model.fit(X_train, y_train)
return model.score(X_test, y_test)
def find_best_c1(args):
set_global_seeds(args['seed'])
dataset = DataLoader(args['dataset'], args)
X_train, X_test, X_val, y_train, y_test, y_val = dataset.prepare_train_test_val(
args)
results = []
for c1 in CLASS_WEIGHTS:
model = SVC(gamma='auto', class_weight={0: 1., 1: c1})
model.fit(X_train, y_train)
results.append(model.score(X_val, y_val))
return results
def find_best_margin(args):
""" return `best_margin / 0.1` """
set_global_seeds(args['seed'])
dataset = DataLoader(args['dataset'])
X_train, X_test, X_val, y_train, y_test, y_val = dataset.prepare_train_test_val(
args)
results = []
for margin in MARGINS:
model = Perceptron(feature_dim=X_train.shape[-1], margin=margin)
model.fit(X_train, y_train)
results.append(model.score(X_val, y_val))
return results
