def train(train_data,
net,
loss,
ctx,
global_step,
epoch_step,
num_epochs,
best_F1=0):
print("Start training on ", ctx)
if isinstance(ctx, mx.Context):
ctx = [ctx]
for epoch in range(num_epochs):
if epoch == 0:
learning_rate = 0.001
print('learning_rate=', learning_rate)
last_loss = None
now_loss = None
trainer = gluon.Trainer(net.collect_params(), 'adam', {
'learning_rate': learning_rate,
'wd': 1e-3
})
else:
different = now_loss - last_loss
if different > 0:
learning_rate = learning_rate * 0.3
trainer = gluon.Trainer(net.collect_params(), 'sgd', {
'learning_rate': learning_rate,
'wd': 1e-3
})
print('learning_rate=', learning_rate)
train_loss, n, = 0.0, 0.0
start = time()
for i, batch in enumerate(train_data):
data, label, batch_size = get_batch(batch, ctx)
losses = []
with autograd.record():
outputs = [net(X) for X in data]
losses = [loss(yhat, y) for yhat, y in zip(outputs, label)]
for l in losses:
l.backward()
sw.add_scalar(tag='cross_entropy',
value=l.mean().asscalar(),
global_step=global_step)
global_step += 1
train_loss += sum([l.sum().asscalar() for l in losses])
n += batch_size
trainer.step(batch_size)
acc_list, tpr_list, tnr_list, ppv_list, ACC, TPR, TNR, PPV, F1 = tool.evaluate(
img_path,
label_path,
net,
crop_size=256,
stride=128,
ctx=ctx[0],
threshold=0.5)
sw.add_scalar(tag='test_acc', value=ACC, global_step=epoch_step)
sw.add_scalar(tag='test_TPR', value=TPR, global_step=epoch_step)
sw.add_scalar(tag='test_TNR', value=TNR, global_step=epoch_step)
sw.add_scalar(tag='test_PPV', value=PPV, global_step=epoch_step)
sw.add_scalar(tag='F1', value=F1, global_step=epoch_step)
epoch_step += 1
print('EPOCH', epoch)
if F1 > best_F1:
net.save_parameters('HEI_fold_4.params')
best_F1 = F1
print('train_loss=', train_loss / n)
if epoch == 0:
last_loss = train_loss / n
now_loss = train_loss / n
else:
last_loss = now_loss
now_loss = train_loss / n
print('time:', time() - start)
sw.close()
def train(train_data,
net,
loss,
ctx,
global_step,
epoch_step,
num_epochs,
best_F1=0):
print("Start training on ", ctx)
if isinstance(ctx, mx.Context):
ctx = [ctx]
for epoch in range(num_epochs):
if epoch == 0:
learning_rate = 0.001
print('learning_rate=', learning_rate)
last_loss = None
now_loss = None
trainer = gluon.Trainer(net.collect_params(), 'adam', {
'learning_rate': learning_rate,
'wd': 1e-3
})
else:
different = now_loss - last_loss
if different > 0:
learning_rate = learning_rate * 0.3
trainer = gluon.Trainer(net.collect_params(), 'sgd', {
'learning_rate': learning_rate,
'wd': 1e-3
})
print('learning_rate=', learning_rate)
train_loss, n, = 0.0, 0.0
start = time()
for i, batch in enumerate(train_data):
data, label, batch_size = get_batch(batch, ctx)
losses = []
with autograd.record():
outputs = [net(X) for X in data]
losses = [loss(yhat, y) for yhat, y in zip(outputs, label)]
for l in losses:
l.backward()
sw.add_scalar(tag='cross_entropy',
value=l.mean().asscalar(),
global_step=global_step)
global_step += 1
train_loss += sum([l.sum().asscalar() for l in losses])
n += batch_size
trainer.step(batch_size)
acc_list, tpr_list, tnr_list, ppv_list, ACC, TPR, TNR, PPV, F1 = tool.evaluate(
img_path,
label_path,
net,
crop_size=256,
stride=128,
ctx=ctx[0],
threshold=0.5)
"""
TP,TN,FP,FN=0,0,0,0
for data,label in test_data:
data=data.as_in_context(ctx[0])
label=label.as_in_context(ctx[0])
pred=net(data)
nd.waitall()
pred=nd.sigmoid(pred)
pred=(pred>0.5).reshape(-1,256,256)
TPt=nd.sum(pred*label).asscalar()
FPt=nd.sum(pred-(pred*label)).asscalar()
FNt=nd.sum(label-(pred*label)).asscalar()
TNt=nd.sum((1-pred)*(1-label)).asscalar()
TP=TP+TPt
FP=FP+FPt
FN=FN+FNt
TN=TN+TNt
ACC=(TP+TN)/(TP+TN+FP+FN+1e-15)
TPR=TP/ (TP+ FN+1e-15)
TNR= TN/(FP+TN+1e-15)
PPV=TP/(TP+FP+1e-15)
F1=2*PPV*TPR/(PPV+TPR+1e-15)
print("acc=",ACC)
print('TPR=',TPR)
print('TNR=',TNR)
print('PPV=',PPV)
print('F1=',F1)
"""
sw.add_scalar(tag='test_acc', value=ACC, global_step=epoch_step)
sw.add_scalar(tag='test_TPR', value=TPR, global_step=epoch_step)
sw.add_scalar(tag='test_TNR', value=TNR, global_step=epoch_step)
sw.add_scalar(tag='test_PPV', value=PPV, global_step=epoch_step)
sw.add_scalar(tag='F1', value=F1, global_step=epoch_step)
epoch_step += 1
print('EPOCH', epoch)
if F1 > best_F1:
net.save_parameters('HEI_fold_1.params')
best_F1 = F1
print('train_loss=', train_loss / n)
if epoch == 0:
last_loss = train_loss / n
now_loss = train_loss / n
else:
last_loss = now_loss
now_loss = train_loss / n
print('time:', time() - start)
sw.close()