(cfg.train_x, cfg.train_y), (cfg.valid_x, cfg.valid_y),
batch_size=cfg.batch_size)
log('> ... finetuning the model')
while (cfg.lrate.get_rate() != 0):
# one epoch of sgd training
train_error, pred, labels = train_sgd(train_fn, cfg)
n_data_train = len(pred)
log("-->> Epoch %d, \n " % cfg.lrate.epoch)
log("- Training:\n")
if multi_label:
format_results(train_error, pred, labels, multi_label, cfg)
else:
format_results(train_error, pred, cfg.train_sets.label_vec, multi_label, cfg)
# validation
valid_error, pred2, labels = validate_by_minibatch(valid_fn, cfg)
n_data_valid = len(pred2)
log("- Validation:\n")
if multi_label:
format_results(valid_error, pred2, labels, multi_label, cfg)
else:
format_results(valid_error, pred2, cfg.valid_sets.label_vec, multi_label, cfg)
cfg.lrate.get_next_rate(current_error = 100 * numpy.mean(valid_error))
# output nnet parameters and lrate, for training resume
if cfg.lrate.epoch % cfg.model_save_step == 0:
_nnet2file(dnn.layers, filename=wdir + '/nnet.tmp')
_lrate2file(cfg.lrate, wdir + '/training_state.tmp')
# save the model and network configuration
for i in range(len(cfg.hidden_layers_sizes)):
if i==0:
dnn = CLDNNV(numpy_rng=numpy_rng, theano_rng=theano_rng, cfg=cfg)
# get the training, validation and testing function for the model
log('> ... getting the finetuning functions')
train_fn, valid_fn = dnn.build_finetune_functions(
(cfg.train_x, cfg.train_y), (cfg.valid_x, cfg.valid_y),
batch_size=cfg.batch_size)
log('> ... finetuning the model')
while (cfg.lrate.get_rate() != 0):
# one epoch of sgd training
train_error = train_sgd(train_fn, cfg)
log('> epoch %d, training error %0.4f ' %
(cfg.lrate.epoch, numpy.mean(train_error)))
# validation
valid_error = validate_by_minibatch(valid_fn, cfg)
log('> epoch %d, lrate %f, validation error %0.4f ' %
(cfg.lrate.epoch, cfg.lrate.get_rate(), numpy.mean(valid_error)))
cfg.lrate.get_next_rate(current_error=numpy.mean(valid_error))
# output nnet parameters and lrate, for training resume
if cfg.lrate.epoch % cfg.model_save_step == 0:
_nnet2file(dnn.layers, filename=wdir + '/nnet.tmp')
_lrate2file(cfg.lrate, wdir + '/training_state.tmp')
# save the model and network configuration
if cfg.param_output_file != '':
_nnet2file(dnn.layers,
filename=cfg.param_output_file,
input_factor=cfg.input_dropout_factor,
factor=cfg.dropout_factor)
log('> ... the final PDNN model parameter is ' + cfg.param_output_file)
if resume_training:
_file2nnet(dnn.layers, filename = wdir + '/nnet.tmp')
# get the training, validation and testing function for the model
log('> ... getting the finetuning functions')
train_fn, valid_fn = dnn.build_finetune_functions(
(cfg.train_x, cfg.train_y), (cfg.valid_x, cfg.valid_y),
batch_size=cfg.batch_size)
log('> ... finetuning the model')
while (cfg.lrate.get_rate() != 0):
# one epoch of sgd training
train_error = train_sgd(train_fn, cfg)
log('> epoch %d, training error %f ' % (cfg.lrate.epoch, 100*numpy.mean(train_error)) + '(%)')
# validation
valid_error = validate_by_minibatch(valid_fn, cfg)
log('> epoch %d, lrate %f, validation error %f ' % (cfg.lrate.epoch, cfg.lrate.get_rate(), 100*numpy.mean(valid_error)) + '(%)')
cfg.lrate.get_next_rate(current_error = 100*numpy.mean(valid_error))
# output nnet parameters and lrate, for training resume
if cfg.lrate.epoch % cfg.model_save_step == 0:
_nnet2file(dnn.layers, filename=wdir + '/nnet.tmp')
_lrate2file(cfg.lrate, wdir + '/training_state.tmp')
# save the model and network configuration
if cfg.param_output_file != '':
_nnet2file(dnn.layers, filename=cfg.param_output_file, input_factor = cfg.input_dropout_factor, factor = cfg.dropout_factor)
log('> ... the final PDNN model parameter is ' + cfg.param_output_file)
if cfg.cfg_output_file != '':
_cfg2file(dnn.cfg, filename=cfg.cfg_output_file)
log('> ... the final PDNN model config is ' + cfg.cfg_output_file)
for n in active_tasks:
if batch_index < batch_numbers_per_chunk[n]:
train_error_array[n].append(train_fn_array[n](index=batch_index, learning_rate = config_array[n].lrate.get_rate(), momentum = config_array[n].momentum))
# now check whether we finish one epoch on any of the tasks
active_tasks_new = active_tasks
for n in active_tasks:
cfg = config_array[n]
if cfg.train_sets.is_finish(): # if true, we reach the end of one epoch of task #n
# reset data reading to the start of next epoch, and output the training error
cfg.train_sets.initialize_read()
if n == 0:
log('> task %d, epoch %d, training error %f ' % (n, cfg.lrate.epoch, 100*numpy.mean(train_error_array[n])) + '(%)')
train_error_array[n] = []
# perform validation, output valid error rate, and adjust learning rate based on the learning rate
valid_error = validate_by_minibatch(valid_fn_array[n], cfg)
log('> task %d, epoch %d, lrate %f, validation error %f ' % (n, cfg.lrate.epoch, cfg.lrate.get_rate(), 100*numpy.mean(valid_error)) + '(%)')
cfg.lrate.get_next_rate(current_error = 100 * numpy.mean(valid_error))
else:
log('> task %d, epoch %d, training error %f ' % (n, cfg.lrate.epoch, numpy.mean(train_error_array[n])) + '(%)')
train_error_array[n] = []
# perform validation, output valid error rate, and adjust learning rate based on the learning rate
valid_error = validate_by_minibatch(valid_fn_array[n], cfg)
log('> task %d, epoch %d, lrate %f, validation error %f ' % (n, cfg.lrate.epoch, cfg.lrate.get_rate(), numpy.mean(valid_error)) + '(%)')
cfg.lrate.get_next_rate(current_error = numpy.mean(valid_error))
# output nnet parameters and lrate, for training resume
_nnet2file(dnn_array[n].layers, filename=wdir + '/nnet.tmp.task' + str(n))
_lrate2file(cfg.lrate, wdir + '/training_state.tmp.task' + str(n))
# if the lrate of a task decays to 0, training on this task terminates; it will be excluded from future training
if cfg.lrate.get_rate() == 0:
active_tasks_new.remove(n)
correct_number = 0.0
confusion_matrix = numpy.zeros((cfg.n_outs, cfg.n_outs))
class_occurrence = numpy.zeros((1,cfg.n_outs))
for i in range(len(pred)):
p_sorted = pred[i]
if p_sorted == labels[i]:
correct_number += 1
confusion_matrix[labels[i], labels[i]] += 1
else:
confusion_matrix[labels[i], p_sorted] += 1
class_occurrence[0, labels[i]] += 1
confusion_matrix = 100 * confusion_matrix / class_occurrence.T
log('-->> Epoch %d, training error %f ' % (cfg.lrate.epoch, 100 * numpy.mean(train_error)) + '(%)')
log('Confusion Matrix is \n\n ' + str(numpy.around(confusion_matrix, 2)) + ' (%)\n')
# validation
valid_error, pred2 = validate_by_minibatch(valid_fn, cfg)
labels = cfg.valid_sets.label_vec
correct_number = 0.0
confusion_matrix = numpy.zeros((cfg.n_outs, cfg.n_outs))
class_occurrence = numpy.zeros((1, cfg.n_outs))
for i in range(len(pred2)):
p_sorted = pred2[i]
if p_sorted == labels[i]:
correct_number += 1
confusion_matrix[labels[i], labels[i]] += 1
else:
confusion_matrix[labels[i], p_sorted] += 1
class_occurrence[0, labels[i]] += 1
confusion_matrix = 100 * confusion_matrix / class_occurrence.T
error_rate = 100 * (1.0 - correct_number / pred2.shape[0])
log('-->> Epoch %d, lrate %f, validation error %f ' % (cfg.lrate.epoch, cfg.lrate.get_rate(),
_file2nnet(dnn, filename = wdir + '/nnet.tmp_DNN')
# get the training, validation and testing function for the model
log('> ... getting the finetuning functions')
train_fn, valid_fn = dnn.build_finetune_functions(
(cfg.train_x, cfg.train_y), (cfg.valid_x, cfg.valid_y),
batch_size=cfg.batch_size)
log('> ... finetuning the model')
count = 0
while (cfg.lrate.get_rate() != 0):
# one epoch of sgd training
train_acc, train_err = train_sgd(train_fn, cfg)
log('> epoch %d, lrate %f, training accuracy %f' % (cfg.lrate.epoch, cfg.lrate.get_rate(), 100*numpy.mean(train_acc)) + '(%)')
# validation
valid_acc, valid_err = validate_by_minibatch(valid_fn, cfg)
log('> epoch %d, lrate %f, validate accuracy %f' % (cfg.lrate.epoch, cfg.lrate.get_rate(), 100*numpy.mean(valid_acc)) + '(%)')
cfg.lrate.get_next_rate(current_error = 100*numpy.mean(valid_err))
# output nnet parameters and lrate, for training resume
if cfg.lrate.epoch % cfg.model_save_step == 0:
_nnet2file(dnn, filename=wdir + '/nnet.tmp_DNN')
_lrate2file(cfg.lrate, wdir + '/training_state.tmp_DNN')
# save the model and network configuration
if cfg.param_output_file != '':
_nnet2file(dnn, filename=cfg.param_output_file, input_factor = cfg.input_dropout_factor, factor = cfg.dropout_factor)
log('> ... the final PDNN model parameter is ' + cfg.param_output_file)
if cfg.cfg_output_file != '':
_cfg2file(dnn.cfg, filename=cfg.cfg_output_file)
log('> ... the final PDNN model config is ' + cfg.cfg_output_file)
def dnn_run(arguments):
required_arguments = ['train_data', 'valid_data', 'nnet_spec', 'wdir']
for arg in required_arguments:
if arguments.has_key(arg) == False:
print "Error: the argument %s has to be specified" % (arg); exit(1)
train_data_spec = arguments['train_data']
valid_data_spec = arguments['valid_data']
nnet_spec = arguments['nnet_spec']
wdir = arguments['wdir']
cfg = NetworkConfig()
cfg.parse_config_dnn(arguments, nnet_spec)
cfg.init_data_reading(train_data_spec, valid_data_spec)
# parse pre-training options
# pre-training files and layer number (how many layers are set to the pre-training parameters)
ptr_layer_number = 0; ptr_file = ''
if arguments.has_key('ptr_file') and arguments.has_key('ptr_layer_number'):
ptr_file = arguments['ptr_file']
ptr_layer_number = int(arguments['ptr_layer_number'])
# check working dir to see whether it's resuming training
resume_training = False
if os.path.exists(wdir + '/nnet.tmp') and os.path.exists(wdir + '/training_state.tmp'):
resume_training = True
cfg.lrate = _file2lrate(wdir + '/training_state.tmp')
log('> ... found nnet.tmp and training_state.tmp, now resume training from epoch ' + str(cfg.lrate.epoch))
numpy_rng = numpy.random.RandomState(89677)
theano_rng = RandomStreams(numpy_rng.randint(2 ** 30))
log('> ... building the model')
# setup model
if cfg.do_dropout:
dnn = DNN_Dropout(numpy_rng=numpy_rng, theano_rng = theano_rng, cfg = cfg)
else:
dnn = DNN(numpy_rng=numpy_rng, theano_rng = theano_rng, cfg = cfg)
# initialize model parameters
# if not resuming training, initialized from the specified pre-training file
# if resuming training, initialized from the tmp model file
if (ptr_layer_number > 0) and (resume_training is False):
_file2nnet(dnn.layers, set_layer_num = ptr_layer_number, filename = ptr_file)
if resume_training:
_file2nnet(dnn.layers, filename = wdir + '/nnet.tmp')
# get the training, validation and testing function for the model
log('> ... getting the finetuning functions')
train_fn, valid_fn = dnn.build_finetune_functions((cfg.train_x, cfg.train_y), (cfg.valid_x, cfg.valid_y),
batch_size=cfg.batch_size)
log('> ... finetuning the model')
while (cfg.lrate.get_rate() != 0):
# one epoch of sgd training
train_error = train_sgd(train_fn, cfg)
log('> epoch %d, training error %f ' % (cfg.lrate.epoch, 100*numpy.mean(train_error)) + '(%)')
# validation
valid_error = validate_by_minibatch(valid_fn, cfg)
log('> epoch %d, lrate %f, validation error %f ' % (cfg.lrate.epoch, cfg.lrate.get_rate(), 100*numpy.mean(valid_error)) + '(%)')
cfg.lrate.get_next_rate(current_error = 100*numpy.mean(valid_error))
# output nnet parameters and lrate, for training resume
if cfg.lrate.epoch % cfg.model_save_step == 0:
_nnet2file(dnn.layers, filename=wdir + '/nnet.tmp')
_lrate2file(cfg.lrate, wdir + '/training_state.tmp')
# save the model and network configuration
if cfg.param_output_file != '':
_nnet2file(dnn.layers, filename=cfg.param_output_file, input_factor = cfg.input_dropout_factor, factor = cfg.dropout_factor)
log('> ... the final PDNN model parameter is ' + cfg.param_output_file)
if cfg.cfg_output_file != '':
_cfg2file(dnn.cfg, filename=cfg.cfg_output_file)
log('> ... the final PDNN model config is ' + cfg.cfg_output_file)
def dnn_run(arguments):
required_arguments = ['train_data', 'valid_data', 'nnet_spec', 'wdir']
for arg in required_arguments:
if arguments.has_key(arg) == False:
print "Error: the argument %s has to be specified" % (arg)
exit(1)
train_data_spec = arguments['train_data']
valid_data_spec = arguments['valid_data']
nnet_spec = arguments['nnet_spec']
wdir = arguments['wdir']
cfg = NetworkConfig()
cfg.parse_config_dnn(arguments, nnet_spec)
cfg.init_data_reading(train_data_spec, valid_data_spec)
# parse pre-training options
# pre-training files and layer number (how many layers are set to the pre-training parameters)
ptr_layer_number = 0
ptr_file = ''
if arguments.has_key('ptr_file') and arguments.has_key('ptr_layer_number'):
ptr_file = arguments['ptr_file']
ptr_layer_number = int(arguments['ptr_layer_number'])
# check working dir to see whether it's resuming training
resume_training = False
if os.path.exists(wdir +
'/nnet.tmp') and os.path.exists(wdir +
'/training_state.tmp'):
resume_training = True
cfg.lrate = _file2lrate(wdir + '/training_state.tmp')
log('> ... found nnet.tmp and training_state.tmp, now resume training from epoch '
+ str(cfg.lrate.epoch))
numpy_rng = numpy.random.RandomState(89677)
theano_rng = RandomStreams(numpy_rng.randint(2**30))
log('> ... building the model')
# setup model
if cfg.do_dropout:
dnn = DNN_Dropout(numpy_rng=numpy_rng, theano_rng=theano_rng, cfg=cfg)
else:
dnn = DNN(numpy_rng=numpy_rng, theano_rng=theano_rng, cfg=cfg)
# initialize model parameters
# if not resuming training, initialized from the specified pre-training file
# if resuming training, initialized from the tmp model file
if (ptr_layer_number > 0) and (resume_training is False):
_file2nnet(dnn.layers,
set_layer_num=ptr_layer_number,
filename=ptr_file)
if resume_training:
_file2nnet(dnn.layers, filename=wdir + '/nnet.tmp')
# get the training, validation and testing function for the model
log('> ... getting the finetuning functions')
train_fn, valid_fn = dnn.build_finetune_functions(
(cfg.train_x, cfg.train_y), (cfg.valid_x, cfg.valid_y),
batch_size=cfg.batch_size)
log('> ... finetuning the model')
while (cfg.lrate.get_rate() != 0):
# one epoch of sgd training
train_error = train_sgd(train_fn, cfg)
log('> epoch %d, training error %f ' %
(cfg.lrate.epoch, 100 * numpy.mean(train_error)) + '(%)')
# validation
valid_error = validate_by_minibatch(valid_fn, cfg)
log('> epoch %d, lrate %f, validation error %f ' %
(cfg.lrate.epoch, cfg.lrate.get_rate(),
100 * numpy.mean(valid_error)) + '(%)')
cfg.lrate.get_next_rate(current_error=100 * numpy.mean(valid_error))
# output nnet parameters and lrate, for training resume
if cfg.lrate.epoch % cfg.model_save_step == 0:
_nnet2file(dnn.layers, filename=wdir + '/nnet.tmp')
_lrate2file(cfg.lrate, wdir + '/training_state.tmp')
# save the model and network configuration
if cfg.param_output_file != '':
_nnet2file(dnn.layers,
filename=cfg.param_output_file,
input_factor=cfg.input_dropout_factor,
factor=cfg.dropout_factor)
log('> ... the final PDNN model parameter is ' + cfg.param_output_file)
if cfg.cfg_output_file != '':
_cfg2file(dnn.cfg, filename=cfg.cfg_output_file)
log('> ... the final PDNN model config is ' + cfg.cfg_output_file)