@author: brummli
"""
import baseParser
import config_
from experiment import AutoencoderGWRExperiment
import nn_main
import gwr_main
def addExtraGWRArguments(parser):
parser.add_argument("-gle", "--GWR_lengthExample", type=baseParser.check_positive, help='The number of frames to use', default=1) #TODO: change default
parser.add_argument("-ghe", "--GWR_hopExample", type=baseParser.check_positive, help='The number of frames to advance per example', default=1) #TODO: change default
parser.add_argument("-gmo", "--GWR_NNModelName", type=str, help='Specifies the feature extraction neural net by its file name', default=None)
parser.add_argument("-el", "--extractionLayer", type=baseParser.check_positive_zero, help='Which layers output of the autoencoder will be used to extract a transformed representation', default=config_.extractionLayer)
parser.set_defaults(modelType='GWR')
return parser
if __name__ == '__main__':
parser = baseParser.createBaseParser()
nnParser = nn_main.addNNArguments(parser)
gwrParser = gwr_main.addGWRArguments(nnParser)
combinedParser = addExtraGWRArguments(gwrParser)
fullParser = baseParser.addModeParsers(combinedParser)
args = fullParser.parse_args()
exp = AutoencoderGWRExperiment(args)
args.func(args,exp)
type=float,
help='Bla',
default=config_.epsilonB)
GWRGroup.add_argument("-en",
"--epsilonN",
type=float,
help='Bla',
default=config_.epsilonN)
GWRGroup.add_argument("-tb",
"--tauB",
type=float,
help="tau B",
default=config_.tauB)
GWRGroup.add_argument("-tn",
"--tauN",
type=float,
help="tau N",
default=config_.tauN)
parser.set_defaults(modelType='GWR')
return parser
if __name__ == '__main__':
parser = baseParser.createBaseParser()
gwrParser = addGWRArguments(parser)
fullParser = baseParser.addModeParsers(gwrParser)
args = fullParser.parse_args()
exp = GWRExperiment(args)
args.func(args, exp)
# -*- coding: utf-8 -*-
"""
Created on Mon Feb 19 13:21:06 2018
@author: brummli
"""
import baseParser
import config_
from experiment import KMeansExperiment
def addKMeansArguments(parser):
KMeansGroup = parser.add_argument_group('KMeans', 'K-Means specific values')
KMeansGroup.add_argument("-ncl", "--numCluster", type=baseParser.check_positive, help='Number of cluster for K-Means', default=config_.numCluster)
KMeansGroup.add_argument("-pc", "--patience", type=baseParser.check_positive_zero, help='Number of iterations without improvement before stopping training', default=config_.patience)
KMeansGroup.add_argument("-re", "--reassignment", type=float, help='Fraction when to reassign cluster centers', default=config_.reassignment) #TODO:description
KMeansGroup.add_argument("-tol", "--tolerance", type=float, help='Improvement within tolerance is accepted as converged', default=config_.KMeans_tolerance)
#KMeansGroup.add_argument("-b", "--batch_size", type=check_positive, help='The batch size used during training', default=config_.KMeans_batch_size) TODO: when we use separate files or parsers per algorithm
parser.set_defaults(modelType='KMeans')
return parser
if __name__ == '__main__':
parser = baseParser.createBaseParser()
kmeansParser = addKMeansArguments(parser)
fullParser = baseParser.addModeParsers(kmeansParser)
args = fullParser.parse_args()
exp = KMeansExperiment(args)
args.func(args,exp)
# -*- coding: utf-8 -*-
"""
Created on Mon Feb 19 13:21:06 2018
@author: brummli
"""
import baseParser
import config_
from experiment import NNExperiment
def addNNArguments(parser):
nnGroup = parser.add_argument_group('NN', 'Neural Net specific values')
nnGroup.add_argument("-l", "--layers", type=baseParser.check_positive_zero, help='Number of layers in the neural net model', default=config_.layers)
nnGroup.add_argument("-hid", "--hiddenNodes", type=baseParser.check_positive, help='Number of hidden Units in each layer', default=config_.hiddenNodes)
nnGroup.add_argument("-lr", "--learningRate", type=float, help='Initial learning rate', default=config_.lr)
nnGroup.add_argument("-s", "--sigma", type=float, help='Width of gaussian noise added for denoising autoencoder', default=config_.sigma)
parser.add_argument("modelType", help='defines the model type',choices=['RNN_AE','LSTM','FF','FF_AE','SEQ_AE','SEQ'])
return parser
if __name__ == '__main__':
parser = baseParser.createBaseParser()
nnParser = addNNArguments(parser)
fullParser = baseParser.addModeParsers(nnParser)
args = fullParser.parse_args()
exp = NNExperiment(args)
args.func(args,exp)