def create_model():
from elektronn2 import neuromancer
import theano.tensor as T
import numpy as np
in_sh = (None,1,572-32*9,572-32*9)
img = neuromancer.Input(in_sh, 'b,f,x,y', name='raw')
out0 = neuromancer.Conv(img, 64, (3,3), (1,1))
out1 = neuromancer.Conv(out0, 64, (3,3), (1,1))
out2 = neuromancer.Pool(out1, (2,2))
out3 = neuromancer.Conv(out2, 128, (3,3), (1,1))
out4 = neuromancer.Conv(out3, 128, (3,3), (1,1))
out5 = neuromancer.Pool(out4, (2,2))
out6 = neuromancer.Conv(out5, 256, (3,3), (1,1))
out7 = neuromancer.Conv(out6, 256, (3,3), (1,1))
out8 = neuromancer.Pool(out7, (2,2))
out9 = neuromancer.Conv(out8, 512, (3,3), (1,1))
out10 = neuromancer.Conv(out9, 512, (3,3), (1,1))
out11 = neuromancer.Pool(out10, (2,2))
out12 = neuromancer.Conv(out11, 1024, (3,3), (1,1))
out13 = neuromancer.Conv(out12, 1024, (3,3), (1,1))
out14 = neuromancer.Pool(out13, (2,2))
####
up0 = neuromancer.UpConvMerge(out10, out14, 1024)
up1 = neuromancer.Conv(up0, 512, (3,3), (1,1))
up2 = neuromancer.Conv(up1, 512, (3,3), (1,1))
up3 = neuromancer.UpConvMerge(out7, up2, 512)
up4 = neuromancer.Conv(up3, 256, (3,3), (1,1))
up5 = neuromancer.Conv(up4, 256, (3,3), (1,1))
up6 = neuromancer.UpConvMerge(out4, up5, 256)
up7 = neuromancer.Conv(up6, 128, (3,3), (1,1))
up8 = neuromancer.Conv(up7, 128, (3,3), (1,1))
up9 = neuromancer.UpConvMerge(out1, up8, 128)
up10 = neuromancer.Conv(up9, 64, (3,3), (1,1))
top_feat = neuromancer.Conv(up10, 64, (3,3), (1,1))
# Target outputs
barr_out = neuromancer.Conv(top_feat, 3, (1,1), (1,1), activation_func='lin', name='barr')
obj_out = neuromancer.Conv(top_feat, 4, (1,1), (1,1), activation_func='lin', name='obj')
my_out = neuromancer.Conv(top_feat, 3, (1,1), (1,1), activation_func='lin', name='my')
barr_out = neuromancer.Softmax(barr_out)
obj_out = neuromancer.Softmax(obj_out)
my_out = neuromancer.Softmax(my_out)
target = neuromancer.Input_like(top_feat, dtype='int16', override_f=3, name='target')
barr, obj, my = neuromancer.split(target, 'f', n_out=3, name=['barr_t', 'obj_t', 'my_t'])
# Target loss
barr_loss_pix = neuromancer.MultinoulliNLL(barr_out, barr, target_is_sparse=True,name='nll_barr')
obj_loss_pix = neuromancer.MultinoulliNLL(obj_out, obj, target_is_sparse=True, name='nll_obj')
my_loss_pix = neuromancer.MultinoulliNLL(my_out, my, target_is_sparse=True, name='nll_my')
pred = neuromancer.Concat([barr_out, obj_out, my_out], axis='f')
pred.feature_names = ['barrier_bg', 'barr_mem', 'barr_ecs', 'obj_bg',
'obj_mito', 'obj_ves', 'obj_syn', 'my_bg', 'my_out', 'my_in']
# Objective
weights = np.array([2.154, 0.42, 0.42])
weights *= len(weights) / weights.sum()
loss = neuromancer.AggregateLoss([barr_loss_pix,
obj_loss_pix,
my_loss_pix],
mixing_weights=weights)
# Monitoring / Debug outputs
nll_barr = neuromancer.ApplyFunc(barr_loss_pix, T.mean, name='mnll_barr')
nll_obj = neuromancer.ApplyFunc(obj_loss_pix, T.mean, name='mnll_obj')
nll_my = neuromancer.ApplyFunc(my_loss_pix, T.mean, name='mnll_my')
errors = neuromancer.Errors(barr_out, barr, target_is_sparse=True)
model = neuromancer.model_manager.getmodel()
model.designate_nodes(input_node=img, target_node=target, loss_node=loss,
prediction_node=pred,
prediction_ext=[loss, errors, pred],
debug_outputs =[nll_barr, errors, nll_obj, nll_my])
return model
def create_model():
from elektronn2 import neuromancer
act = 'relu'
in_sh = (batch_size, 4, 3, 128, 256)
inp = neuromancer.Input(in_sh, 'b,f,z,y,x', name='raw')
out0 = neuromancer.Conv(inp,
13, (1, 5, 5), (1, 2, 2),
activation_func=act,
dropout_rate=dr)
out0 = neuromancer.Conv(out0,
19, (1, 5, 5), (1, 2, 2),
activation_func=act,
dropout_rate=dr)
out0 = neuromancer.Conv(out0,
25, (1, 4, 4), (1, 2, 2),
activation_func=act,
dropout_rate=dr)
out0 = neuromancer.Conv(out0,
25, (1, 4, 4), (1, 2, 2),
activation_func=act,
dropout_rate=dr)
out0 = neuromancer.Conv(out0,
30, (1, 2, 2), (1, 2, 2),
activation_func=act,
dropout_rate=dr)
out0 = neuromancer.Conv(out0,
30, (1, 1, 1), (1, 2, 2),
activation_func=act,
dropout_rate=dr)
out = neuromancer.Conv(out0,
31, (1, 1, 1), (1, 1, 1),
activation_func=act,
dropout_rate=dr)
out0, out1, out2 = neuromancer.split(out, axis="z", n_out=3)
out0 = neuromancer.Reshape(out0,
shape=(inp.shape[0], out0.shape.stripbatch_prod,
1),
tags="b,f,z")
out1 = neuromancer.Reshape(out1,
shape=(inp.shape[0], out1.shape.stripbatch_prod,
1),
tags="b,f,z")
out2 = neuromancer.Reshape(out2,
shape=(inp.shape[0], out2.shape.stripbatch_prod,
1),
tags="b,f,z")
out = neuromancer.Concat([out0, out1, out2], axis="z")
out = neuromancer.Perceptron(out, 40, flatten=False, dropout_rate=dr)
out = neuromancer.Perceptron(out, 10, flatten=False, dropout_rate=dr)
out0, out1, out2 = neuromancer.split(out, axis="z", n_out=3)
d_small = neuromancer.EuclideanDistance(out0, out1)
d_big = neuromancer.EuclideanDistance(out0, out2)
loss = neuromancer.RampLoss(d_small, d_big, margin=0.2)
loss = neuromancer.AggregateLoss(loss)
model = neuromancer.model_manager.getmodel()
# model = neuromancer.model.modelload("/wholebrain/scratch/pschuber/CNN_Training/nupa_cnn/t_net/ssv6_tripletnet_v9/ssv6_tripletnet_v9-FINAL.mdl")
model.designate_nodes(input_node=inp,
target_node=None,
loss_node=loss,
prediction_node=out,
prediction_ext=[loss, loss, out])
# params = neuromancer.model.params_from_model_file("/wholebrain/scratch/pschuber/CNN_Training/nupa_cnn/t_net/ssv6_tripletnet_v9/ssv6_tripletnet_v9-FINAL.mdl")
# params = dict(filter(lambda x: x[0].startswith('conv'), params.items()))
# model.set_param_values(params)
return model
# if __name__ == "__main__":
# model = create_model()
# "Test" if model is saveable
# model.save("/tmp/"+save_name)
def create_model():
from elektronn2 import neuromancer as nm
trainee = trainee_dict["create_model"]()
inp = trainee.input_node
trainee_gt = trainee.target_node
trainee_out = trainee.prediction_node
trainee_loss = trainee.loss_node
adv_input, adv_target = nm.advmerge(trainee_out, trainee_gt)
# raw data
diff = np.array(inp.shape.spatial_shape, dtype=np.int32) - np.array(
trainee_out.shape.spatial_shape, dtype=np.int32)
assert not np.any(diff % 2)
raw_inp = nm.Crop(inp, list((diff // 2)))
conv0 = nm.Conv(raw_inp,
5, (2, 3, 3),
activation_func=act,
dropout_rate=dr,
name="conv_adv")
conv1 = nm.Conv(conv0,
10, (2, 3, 3),
activation_func=act,
dropout_rate=dr,
name="conv_adv")
down0 = nm.Pool(conv1, (1, 2, 2), mode='max') # mid res
conv2 = nm.Conv(down0,
10, (2, 3, 3),
activation_func=act,
dropout_rate=dr,
name="conv_adv")
conv3 = nm.Conv(conv2,
15, (2, 3, 3),
activation_func=act,
dropout_rate=dr,
name="conv_adv")
# Merging low-res with mid-res features
mrg1 = nm.UpConvMerge(conv1, conv3, 20, name="upconv_adv")
mconv2 = nm.Conv(mrg1,
15, (2, 3, 3),
activation_func=act,
dropout_rate=dr,
name="conv_adv")
raw_out = nm.Conv(mconv2,
15, (2, 3, 3),
activation_func=act,
dropout_rate=dr,
name="conv_adv")
# segmentation
conv0 = nm.Conv(adv_input,
5, (2, 3, 3),
activation_func=act,
dropout_rate=dr,
name="conv_adv")
conv1 = nm.Conv(conv0,
10, (2, 3, 3),
activation_func=act,
dropout_rate=dr,
name="conv_adv")
down0 = nm.Pool(conv1, (1, 2, 2), mode='max') # mid res
conv2 = nm.Conv(down0,
10, (2, 3, 3),
activation_func=act,
dropout_rate=dr,
name="conv_adv")
conv3 = nm.Conv(conv2,
15, (2, 3, 3),
activation_func=act,
dropout_rate=dr,
name="conv_adv")
# Merging low-res with mid-res features
mrg1 = nm.UpConvMerge(conv1, conv3, 20, name="upconv_adv")
mconv2 = nm.Conv(mrg1,
15, (2, 3, 3),
activation_func=act,
dropout_rate=dr,
name="conv_adv")
seg_out = nm.Conv(mconv2,
15, (2, 3, 3),
activation_func=act,
dropout_rate=dr,
name="conv_adv")
out = nm.Concat([raw_out, seg_out], axis="f")
out = nm.Conv(out,
20, (2, 5, 5),
pool_shape=(1, 2, 2),
activation_func=act,
dropout_rate=dr,
name="conv_adv")
out = nm.Conv(out,
40, (2, 3, 3),
pool_shape=(1, 2, 2),
activation_func=act,
dropout_rate=dr,
name="conv_adv")
out = nm.Conv(out,
60, (2, 2, 2),
pool_shape=(1, 2, 2),
activation_func=act,
dropout_rate=dr,
name="conv_adv")
dec = nm.Perceptron(out,
2,
flatten=True,
activation_func='lin',
name="perc_adv")
adv_out = nm.Softmax(dec)
# as in orig. GAN paper of Godfellow et al. only the positive label is taken
# into account (i.e. adversarial network predicts input is ground truth)
# for the adv. prediction loss when training the trainee
# if training the adv. network only the binary cross-entropy of the adv.
# prediction is backpropagated.
adv_loss = nm.MultinoulliNLL(adv_out,
adv_target,
target_is_sparse=True,
name='nll_adversarial',
class_weights=None)
loss = nm.AggregateLoss([adv_loss, trainee_loss],
mixing_weights=None,
name='loss_adversarial')
model = nm.model_manager.getmodel()
model.designate_nodes(
input_node=inp,
target_node=trainee.target_node,
loss_node=loss,
prediction_node=trainee.prediction_node,
prediction_ext=trainee.prediction_ext,
)
return model
test_vals = np.array([1, 2, 3], dtype=np.float32)
skel = DummySkel()
print(f(test_vals, skel))
print(f_grad(test_vals, skel))
theano.printing.pydotprint(f_grad, '/tmp/step_result.svg')
###########################################################################
from elektronn2 import neuromancer
import theano
act = 'tanh'
data = neuromancer.Input((1, 20), 'b,f', name='data')
mem_0 = neuromancer.Input((1, 120), 'b,f', name='mem')
mlp1 = neuromancer.Dot(data, 120, activation_func=act)
join = neuromancer.Concat([mlp1, mem_0])
out = neuromancer.Dot(join, 120, activation_func=act)
out2 = neuromancer.Dot(out, 13, activation_func='lin')
# recurrent = neuromancer.Scan(out, in_memory=mem_0, n_steps=10)
recurrent, out2r = neuromancer.Scan([out, out2],
out_memory=out,
in_memory=mem_0,
n_steps=7)
loss = neuromancer.AggregateLoss(recurrent)
loss = neuromancer.AggregateLoss(out2r)
grad = theano.grad(loss.output,
loss.all_trainable_params.values(),
disconnected_inputs='warn')
recurrent()
def demo_new_gm():
mfp = False
in_sh = (1, 1, 15, 198, 198) if mfp else (1, 1, 25, 171, 171)
inp = neuromancer.Input(in_sh, 'b,f,z,x,y', name='raw')
out = neuromancer.Conv(inp,
20, (1, 6, 6), (1, 1, 1),
mfp=mfp,
batch_normalisation='train')
out = neuromancer.Conv(out, 40, (1, 5, 5), (1, 2, 2), mfp=mfp)
out = neuromancer.Conv(out, 50, (1, 4, 4), (1, 2, 2), mfp=mfp)
out = neuromancer.Conv(out, 80, (1, 4, 4), (1, 1, 1), mfp=mfp)
out = neuromancer.Conv(out, 80, (4, 1, 1), (2, 1, 1),
mfp=mfp) # first z kernel, 2 pool
out = neuromancer.Conv(out, 80, (3, 4, 4), (1, 1, 1), mfp=mfp)
out = neuromancer.Conv(out, 80, (3, 4, 4), (1, 1, 1), mfp=mfp)
out = neuromancer.Conv(out, 100, (2, 4, 4), (1, 1, 1), mfp=mfp)
out = neuromancer.Conv(out, 120, (2, 4, 4), (1, 1, 1), mfp=mfp)
out = neuromancer.Conv(out, 120, (1, 2, 2), (1, 1, 1), mfp=mfp)
out = neuromancer.Conv(out, 120, (1, 1, 1), (1, 1, 1), mfp=mfp)
out1, out2 = neuromancer.split(out, 1, n_out=2)
probs = neuromancer.Conv(out1,
2, (1, 1, 1), (1, 1, 1),
mfp=mfp,
activation_func='lin')
probs = neuromancer.Softmax(probs, name='probs')
discard, mode = neuromancer.split(probs, 1, n_out=2)
concentration = neuromancer.Conv(out2,
1, (1, 1, 1), (1, 1, 1),
mfp=mfp,
activation_func='lin',
name='concentration')
t_sh = probs.shape.copy()
t_sh.updateshape('f', 1)
target = neuromancer.Input_like(t_sh, dtype='float32', name='target')
loss_pix = neuromancer.BetaNLL(mode, concentration, target)
loss = neuromancer.AggregateLoss(loss_pix)
errors = neuromancer.Errors(probs, target, target_is_sparse=True)
prediction = neuromancer.Concat([mode, concentration],
axis=1,
name='prediction')
loss_std = neuromancer.ApplyFunc(loss_pix, T.std)
model = neuromancer.model_manager.getmodel()
model.designate_nodes(input_node=inp,
target_node=target,
loss_node=loss,
prediction_node=prediction,
prediction_ext=[loss, errors, prediction])
### --- ###
model2 = neuromancer.model_manager.newmodel("second")
inp2 = neuromancer.Input(in_sh, 'b,f,z,x,y', name='raw')
out2 = neuromancer.Conv(inp2, 20, (1, 6, 6), (1, 1, 1), mfp=mfp)
out2 = neuromancer.Conv(out2, 40, (1, 5, 5), (1, 2, 2), mfp=mfp)
out2 = neuromancer.Conv(out2, 50, (1, 4, 4), (1, 2, 2), mfp=mfp)
out2 = neuromancer.Conv(out2, 120, (2, 4, 4), (1, 1, 1), mfp=mfp)
out2 = neuromancer.Conv(out2, 120, (1, 2, 2), (1, 1, 1), mfp=mfp)
out2 = neuromancer.Conv(out2, 120, (1, 1, 1), (1, 1, 1), mfp=mfp)
probs2 = neuromancer.Conv(out2,
2, (1, 1, 1), (1, 1, 1),
mfp=mfp,
activation_func='lin')
probs2 = neuromancer.Softmax(probs2, name='probs')
t_sh = probs2.shape.copy()
t_sh.updateshape('f', 1)
target2 = neuromancer.Input_like(t_sh, dtype='float32', name='target')
loss_pix2 = neuromancer.MultinoulliNLL(probs2, target2)
loss2 = neuromancer.AggregateLoss(loss_pix2)
errors2 = neuromancer.Errors(probs2, target2, target_is_sparse=True)
model2.designate_nodes(input_node=inp2,
target_node=target2,
loss_node=loss2,
prediction_node=probs2,
prediction_ext=[loss2, errors2, probs2])
model.save('/tmp/test.pkl')
model2.save('/tmp/test2.pkl')
model2_reloaded = modelload('/tmp/test2.pkl')
model2_reloaded.save('/tmp/test2_reloaded.pkl')
print(neuromancer.model_manager)