def main():
parser = argparse.ArgumentParser(description='BR')
parser.add_argument('im_size', type=int, help='image size')
parser.add_argument('dataset', type=str, help='dataset')
parser.add_argument('c', type=float, help='amount of dropout', default=0.)
parser.add_argument('lr', type=float, help='learning rate')
parser.add_argument('opt', type=str, choices=['sgd', 'adam'])
parser.add_argument('n_hidden',
type=int,
help='number of neurons in hidden layer')
parser.add_argument('-im',
'--imagenet',
action='store_true',
help='use imagenet weights')
parser.add_argument('-pt',
'--pretrained',
action='store_true',
help='Use pretrained VGG features')
parser.add_argument('-name',
type=str,
help='name of experiment',
default='BR')
args = parser.parse_args()
args.name = '_'.join((args.name, args.dataset, str(args.pretrained)))
timestamp = time.strftime("%Y-%m-%d_%H:%M")
logger = loggerClass(args, timestamp)
BR(args, logger, timestamp)
def main():
parser = argparse.ArgumentParser(description='OR')
parser.add_argument('dataset', type=str, help='dataset')
parser.add_argument('g',
type=float,
help='monotonicity constraint hyperparam',
default=0.1)
parser.add_argument('q',
type=float,
help='L2 shrinkage hyperparam',
default=0.)
parser.add_argument('-pt',
'--pretrained',
action='store_true',
help='Use pretrained VGG features')
parser.add_argument('-name',
type=str,
help='name of experiment',
default='OR_FINAL')
args = parser.parse_args()
args.name = '_'.join((args.name, args.dataset))
timestamp = time.strftime("%Y-%m-%d_%H:%M")
logger = loggerClass(args, timestamp)
OR(args, logger)
def main():
parser = argparse.ArgumentParser(
description='Threshold stacking')
parser.add_argument('dataset', type=str, help='dataset')
parser.add_argument('-pt', '--pretrained', action='store_true',
help='Use marginals predicted on pretrained features')
parser.add_argument('-name', type=str, help='name of experiment', default='THRESH_stack')
parser.add_argument('-nl', '--nonlinear', action='store_true', help='use nonlinear model (RF)')
args = parser.parse_args()
args.name = ' '.join((args.name, args.dataset))
timestamp = time.strftime("%Y-%m-%d_%H:%M")
logger = loggerClass(args, timestamp)
thresh_stack(args, logger)
def main():
parser = argparse.ArgumentParser(description='YE2012')
parser.add_argument('dataset', type=str, help='dataset')
parser.add_argument('-pt',
'--pretrained',
action='store_true',
help='Use pretrained VGG features')
parser.add_argument('-name',
type=str,
help='name of experiment',
default='YE2012')
args = parser.parse_args()
args.name = '_'.join((args.name, args.dataset))
timestamp = time.strftime("%Y-%m-%d_%H:%M")
logger = loggerClass(args, timestamp)
ye_et_al(args, logger)
def main():
parser = argparse.ArgumentParser(description='thresholding')
parser.add_argument('dataset', type=str, help='dataset')
parser.add_argument(
'-pt',
'--pretrained',
action='store_true',
help='Load marginals from experiments with pretrained features')
parser.add_argument('-name',
type=str,
help='name of experiment',
default='THRESH')
args = parser.parse_args()
args.name = '_'.join((args.name, args.dataset))
timestamp = time.strftime("%Y-%m-%d_%H:%M")
logger = loggerClass(args, timestamp)
thresholding(args, logger)