wellton/braindecode
1
0
Fork 0
Código Issues Pull Requests Ações Pacotes Projetos Versões Wiki Atividade

put hyperparameters into one place in code

Ver Código-Fonte
Esse commit está contido em:
Robin Tibor Schirrmeister
2018-08-27 14:22:24 +02:00
pai b0f186e3e0
commit b69a930799
2 arquivos alterados com 41 adições e 25 exclusões
Baixar Arquivo de Patch Baixar Arquivo de Diff Expandir todos os arquivos Colapsar todos os arquivos
examples/bcic_iv_2a.py
+20 -12
Ver Arquivo
@@ -28,6 +28,15 @@ log = logging.getLogger(__name__)
def run_exp(data_folder, subject_id, low_cut_hz, model, cuda):
ival = [-500, 4000]
max_epochs = 1600
max_increase_epochs = 160
batch_size = 60
high_cut_hz = 38
factor_new = 1e-3
init_block_size = 1000
valid_set_fraction = 0.2
train_filename = 'A{:02d}T.gdf'.format(subject_id)
test_filename = 'A{:02d}E.gdf'.format(subject_id)
train_filepath = os.path.join(data_folder, train_filename)
@@ -50,12 +59,12 @@ def run_exp(data_folder, subject_id, low_cut_hz, model, cuda):
# lets convert to millvolt for numerical stability of next operations
train_cnt = mne_apply(lambda a: a * 1e6, train_cnt)
train_cnt = mne_apply(
lambda a: bandpass_cnt(a, low_cut_hz, 38, train_cnt.info['sfreq'],
lambda a: bandpass_cnt(a, low_cut_hz, high_cut_hz, train_cnt.info['sfreq'],
filt_order=3,
axis=1), train_cnt)
train_cnt = mne_apply(
lambda a: exponential_running_standardize(a.T, factor_new=1e-3,
init_block_size=1000,
lambda a: exponential_running_standardize(a.T, factor_new=factor_new,
init_block_size=init_block_size,
eps=1e-4).T,
train_cnt)
@@ -64,24 +73,23 @@ def run_exp(data_folder, subject_id, low_cut_hz, model, cuda):
assert len(test_cnt.ch_names) == 22
test_cnt = mne_apply(lambda a: a * 1e6, test_cnt)
test_cnt = mne_apply(
lambda a: bandpass_cnt(a, low_cut_hz, 38, test_cnt.info['sfreq'],
lambda a: bandpass_cnt(a, low_cut_hz, high_cut_hz, test_cnt.info['sfreq'],
filt_order=3,
axis=1), test_cnt)
test_cnt = mne_apply(
lambda a: exponential_running_standardize(a.T, factor_new=1e-3,
init_block_size=1000,
lambda a: exponential_running_standardize(a.T, factor_new=factor_new,
init_block_size=init_block_size,
eps=1e-4).T,
test_cnt)
marker_def = OrderedDict([('Left Hand', [1]), ('Right Hand', [2],),
('Foot', [3]), ('Tongue', [4])])
ival = [-500, 4000]
train_set = create_signal_target_from_raw_mne(train_cnt, marker_def, ival)
test_set = create_signal_target_from_raw_mne(test_cnt, marker_def, ival)
train_set, valid_set = split_into_two_sets(train_set,
first_set_fraction=0.8)
train_set, valid_set = split_into_two_sets(
train_set, first_set_fraction=1-valid_set_fraction)
set_random_seeds(seed=20190706, cuda=cuda)
@@ -100,10 +108,10 @@ def run_exp(data_folder, subject_id, low_cut_hz, model, cuda):
optimizer = optim.Adam(model.parameters())
iterator = BalancedBatchSizeIterator(batch_size=60)
iterator = BalancedBatchSizeIterator(batch_size=batch_size)
stop_criterion = Or([MaxEpochs(1600),
NoDecrease('valid_misclass', 160)])
stop_criterion = Or([MaxEpochs(max_epochs),
NoDecrease('valid_misclass', max_increase_epochs)])
monitors = [LossMonitor(), MisclassMonitor(), RuntimeMonitor()]
examples/bcic_iv_2a_cropped.py
+21 -13
Ver Arquivo
@@ -30,6 +30,16 @@ log = logging.getLogger(__name__)
def run_exp(data_folder, subject_id, low_cut_hz, model, cuda):
ival = [-500, 4000]
input_time_length = 1000
max_epochs = 800
max_increase_epochs = 80
batch_size = 60
high_cut_hz = 38
factor_new = 1e-3
init_block_size = 1000
valid_set_fraction = 0.2
train_filename = 'A{:02d}T.gdf'.format(subject_id)
test_filename = 'A{:02d}E.gdf'.format(subject_id)
train_filepath = os.path.join(data_folder, train_filename)
@@ -52,12 +62,12 @@ def run_exp(data_folder, subject_id, low_cut_hz, model, cuda):
# lets convert to millvolt for numerical stability of next operations
train_cnt = mne_apply(lambda a: a * 1e6, train_cnt)
train_cnt = mne_apply(
lambda a: bandpass_cnt(a, low_cut_hz, 38, train_cnt.info['sfreq'],
lambda a: bandpass_cnt(a, low_cut_hz, high_cut_hz, train_cnt.info['sfreq'],
filt_order=3,
axis=1), train_cnt)
train_cnt = mne_apply(
lambda a: exponential_running_standardize(a.T, factor_new=1e-3,
init_block_size=1000,
lambda a: exponential_running_standardize(a.T, factor_new=factor_new,
init_block_size=init_block_size,
eps=1e-4).T,
train_cnt)
@@ -66,30 +76,28 @@ def run_exp(data_folder, subject_id, low_cut_hz, model, cuda):
assert len(test_cnt.ch_names) == 22
test_cnt = mne_apply(lambda a: a * 1e6, test_cnt)
test_cnt = mne_apply(
lambda a: bandpass_cnt(a, low_cut_hz, 38, test_cnt.info['sfreq'],
lambda a: bandpass_cnt(a, low_cut_hz, high_cut_hz, test_cnt.info['sfreq'],
filt_order=3,
axis=1), test_cnt)
test_cnt = mne_apply(
lambda a: exponential_running_standardize(a.T, factor_new=1e-3,
init_block_size=1000,
lambda a: exponential_running_standardize(a.T, factor_new=factor_new,
init_block_size=init_block_size,
eps=1e-4).T,
test_cnt)
marker_def = OrderedDict([('Left Hand', [1]), ('Right Hand', [2],),
('Foot', [3]), ('Tongue', [4])])
ival = [-500, 4000]
train_set = create_signal_target_from_raw_mne(train_cnt, marker_def, ival)
test_set = create_signal_target_from_raw_mne(test_cnt, marker_def, ival)
train_set, valid_set = split_into_two_sets(train_set,
first_set_fraction=0.8)
train_set, valid_set = split_into_two_sets(
train_set, first_set_fraction=1-valid_set_fraction)
set_random_seeds(seed=20190706, cuda=cuda)
n_classes = 4
n_chans = int(train_set.X.shape[1])
input_time_length=1000
if model == 'shallow':
model = ShallowFBCSPNet(n_chans, n_classes, input_time_length=input_time_length,
final_conv_length=30).create_network()
@@ -112,12 +120,12 @@ def run_exp(data_folder, subject_id, low_cut_hz, model, cuda):
optimizer = optim.Adam(model.parameters())
iterator = CropsFromTrialsIterator(batch_size=60,
iterator = CropsFromTrialsIterator(batch_size=batch_size,
input_time_length=input_time_length,
n_preds_per_input=n_preds_per_input)
stop_criterion = Or([MaxEpochs(800),
NoDecrease('valid_misclass', 80)])
stop_criterion = Or([MaxEpochs(max_epochs),
NoDecrease('valid_misclass', max_increase_epochs)])
monitors = [LossMonitor(), MisclassMonitor(col_suffix='sample_misclass'),
CroppedTrialMisclassMonitor(