Python lasagne.updates() Examples

The following are 30 code examples of lasagne.updates(). You can vote up the ones you like or vote down the ones you don't like, and go to the original project or source file by following the links above each example. You may also want to check out all available functions/classes of the module lasagne , or try the search function .
Example #1
Source File: deep_conv_classification_alt48_luad10in20_brca10.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #2
Source File: deep_conv_classification_alt48maxp_luad10_luad10in20_brca10x1.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #3
Source File: deep_conv_classification_alt45.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #4
Source File: deep_conv_classification_alt48_luad10_luad10in20_brca10x2.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #5
Source File: deep_conv_classification_alt54.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #6
Source File: deep_conv_classification_alt41.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #7
Source File: deep_conv_classification_alt38.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #8
Source File: deep_conv_classification_alt32.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #9
Source File: deep_conv_classification_alt36_deploy.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #10
Source File: deep_conv_classification_alt49.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #11
Source File: deep_conv_classification_alt59.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #12
Source File: deep_conv_classification_alt46.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #13
Source File: deep_conv_classification_alt48.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #14
Source File: deep_conv_classification_alt44.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #15
Source File: deep_conv_classification_alt48_luad10_skcm10_lr0.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #16
Source File: deep_conv_classification_alt62.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #17
Source File: deep_conv_classification_alt48_adeno_prad_t1_heatmap.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #18
Source File: deep_conv_classification_lpatch_alt2.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #19
Source File: deep_conv_classification_alt48_only_skcm_t0.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #20
Source File: deep_conv_classification_alt48maxp_luad10_luad10in20_brca10x2.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #21
Source File: deep_conv_classification_lpatch_alt0.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #22
Source File: deep_conv_classification_alt36.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #23
Source File: deep_conv_classification_alt61.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #24
Source File: deep_conv_classification_alt48_luad10_skcm10_v0.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #25
Source File: deep_conv_classification_alt29.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #26
Source File: deep_conv_classification_alt48_heatmap_only_melanoma.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #27
Source File: deep_conv_classification_lpatch_alt3.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #28
Source File: deep_conv_classification_alt36-sp-cnn.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #29
Source File: deep_conv_classification_alt27.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;
Example #30
Source File: deep_conv_classification_alt28.py    From u24_lymphocyte with BSD 3-Clause "New" or "Revised" License 6 votes vote down vote up 
def make_training_functions(network, new_params, input_var, aug_var, target_var):
    output = lasagne.layers.get_output(network, deterministic=True, batch_norm_use_averages=True, batch_norm_update_averages=False);
    loss = lasagne.objectives.binary_crossentropy(output, target_var).mean();

    deter_output = lasagne.layers.get_output(network, deterministic=True);
    deter_loss = lasagne.objectives.binary_crossentropy(deter_output, target_var).mean();

    params = layers.get_all_params(network, trainable=True);
    updates = lasagne.updates.nesterov_momentum(loss, params, learning_rate=LearningRate, momentum=0.985);
    new_params_updates = lasagne.updates.nesterov_momentum(loss, new_params, learning_rate=LearningRate, momentum=0.985);

    val_fn = theano.function([input_var, aug_var, target_var], [deter_loss, deter_output]);
    train_fn = theano.function([input_var, aug_var, target_var], loss, updates=updates);
    new_params_train_fn = theano.function([input_var, aug_var, target_var], loss, updates=new_params_updates);

    return train_fn, new_params_train_fn, val_fn;