def init_tree(tree: ProtoTree, optimizer, scheduler, device, args: argparse.Namespace):
epoch = 1
mean = 0.5
std = 0.1
# load trained prototree if flag is set
# NOTE: TRAINING FURTHER FROM A CHECKPOINT DOESN'T SEEM TO WORK CORRECTLY. EVALUATING A TRAINED PROTOTREE FROM A CHECKPOINT DOES WORK.
if args.state_dict_dir_tree != '':
if not args.disable_cuda and torch.cuda.is_available():
device = torch.device('cuda:{}'.format(torch.cuda.current_device()))
else:
device = torch.device('cpu')
if args.disable_cuda or not torch.cuda.is_available():
# tree = load_state(args.state_dict_dir_tree, device)
tree = torch.load(args.state_dict_dir_tree+'/model.pth', map_location=device)
else:
tree = torch.load(args.state_dict_dir_tree+'/model.pth')
tree.to(device=device)
try:
epoch = int(args.state_dict_dir_tree.split('epoch_')[-1]) + 1
except:
epoch=args.epochs+1
print("Train further from epoch: ", epoch, flush=True)
optimizer.load_state_dict(torch.load(args.state_dict_dir_tree+'/optimizer_state.pth', map_location=device))
if epoch>args.freeze_epochs:
for parameter in tree._net.parameters():
parameter.requires_grad = True
if not args.disable_derivative_free_leaf_optim:
for leaf in tree.leaves:
leaf._dist_params.requires_grad = False
if os.path.isfile(args.state_dict_dir_tree+'/scheduler_state.pth'):
# scheduler.load_state_dict(torch.load(args.state_dict_dir_tree+'/scheduler_state.pth'))
# print(scheduler.state_dict(),flush=True)
scheduler.last_epoch = epoch - 1
scheduler._step_count = epoch
elif args.state_dict_dir_net != '': # load pretrained conv network
# initialize prototypes
torch.nn.init.normal_(tree.prototype_layer.prototype_vectors, mean=mean, std=std)
#strict is False so when loading pretrained model, ignore the linear classification layer
tree._net.load_state_dict(torch.load(args.state_dict_dir_net+'/model_state.pth'), strict=False)
tree._add_on.load_state_dict(torch.load(args.state_dict_dir_net+'/model_state.pth'), strict=False)
else:
with torch.no_grad():
# initialize prototypes
torch.nn.init.normal_(tree.prototype_layer.prototype_vectors, mean=mean, std=std)
tree._add_on.apply(init_weights_xavier)
return tree, epoch
def eval_fidelity(tree: ProtoTree,
test_loader: DataLoader,
device,
log: Log = None,
progress_prefix: str = 'Fidelity') -> dict:
tree = tree.to(device)
# Keep an info dict about the procedure
info = dict()
# Make sure the model is in evaluation mode
tree.eval()
# Show progress on progress bar
test_iter = tqdm(enumerate(test_loader),
total=len(test_loader),
desc=progress_prefix,
ncols=0)
distr_samplemax_fidelity = 0
distr_greedy_fidelity = 0
# Iterate through the test set
for i, (xs, ys) in test_iter:
xs, ys = xs.to(device), ys.to(device)
# Use the model to classify this batch of input data, with 3 types of routing
out_distr, _ = tree.forward(xs, 'distributed')
ys_pred_distr = torch.argmax(out_distr, dim=1)
out_samplemax, _ = tree.forward(xs, 'sample_max')
ys_pred_samplemax = torch.argmax(out_samplemax, dim=1)
out_greedy, _ = tree.forward(xs, 'greedy')
ys_pred_greedy = torch.argmax(out_greedy, dim=1)
# Calculate fidelity
distr_samplemax_fidelity += torch.sum(
torch.eq(ys_pred_samplemax, ys_pred_distr)).item()
distr_greedy_fidelity += torch.sum(
torch.eq(ys_pred_greedy, ys_pred_distr)).item()
# Update the progress bar
test_iter.set_postfix_str(f'Batch [{i + 1}/{len(test_iter)}]')
del out_distr
del out_samplemax
del out_greedy
distr_samplemax_fidelity = distr_samplemax_fidelity / float(
len(test_loader.dataset))
distr_greedy_fidelity = distr_greedy_fidelity / float(
len(test_loader.dataset))
info['distr_samplemax_fidelity'] = distr_samplemax_fidelity
info['distr_greedy_fidelity'] = distr_greedy_fidelity
log.log_message(
"Fidelity between standard distributed routing and sample_max routing: "
+ str(distr_samplemax_fidelity))
log.log_message(
"Fidelity between standard distributed routing and greedy routing: " +
str(distr_greedy_fidelity))
return info
def eval(tree: ProtoTree,
test_loader: DataLoader,
epoch,
device,
log: Log = None,
sampling_strategy: str = 'distributed',
log_prefix: str = 'log_eval_epochs',
progress_prefix: str = 'Eval Epoch') -> dict:
tree = tree.to(device)
# Keep an info dict about the procedure
info = dict()
if sampling_strategy != 'distributed':
info['out_leaf_ix'] = []
# Build a confusion matrix
cm = np.zeros((tree._num_classes, tree._num_classes), dtype=int)
# Make sure the model is in evaluation mode
tree.eval()
# Show progress on progress bar
test_iter = tqdm(enumerate(test_loader),
total=len(test_loader),
desc=progress_prefix + ' %s' % epoch,
ncols=0)
# Iterate through the test set
for i, (xs, ys) in test_iter:
xs, ys = xs.to(device), ys.to(device)
# Use the model to classify this batch of input data
out, test_info = tree.forward(xs, sampling_strategy)
ys_pred = torch.argmax(out, dim=1)
# Update the confusion matrix
cm_batch = np.zeros((tree._num_classes, tree._num_classes), dtype=int)
for y_pred, y_true in zip(ys_pred, ys):
cm[y_true][y_pred] += 1
cm_batch[y_true][y_pred] += 1
acc = acc_from_cm(cm_batch)
test_iter.set_postfix_str(
f'Batch [{i + 1}/{len(test_iter)}], Acc: {acc:.3f}')
# keep list of leaf indices where test sample ends up when deterministic routing is used.
if sampling_strategy != 'distributed':
info['out_leaf_ix'] += test_info['out_leaf_ix']
del out
del ys_pred
del test_info
info['confusion_matrix'] = cm
info['test_accuracy'] = acc_from_cm(cm)
log.log_message("\nEpoch %s - Test accuracy with %s routing: " %
(epoch, sampling_strategy) + str(info['test_accuracy']))
return info
def run_tree(args=None):
args = args or get_args()
# Create a logger
log = Log(args.log_dir)
print("Log dir: ", args.log_dir, flush=True)
# Create a csv log for storing the test accuracy, mean train accuracy and mean loss for each epoch
log.create_log('log_epoch_overview', 'epoch', 'test_acc', 'mean_train_acc', 'mean_train_crossentropy_loss_during_epoch')
# Log the run arguments
save_args(args, log.metadata_dir)
if not args.disable_cuda and torch.cuda.is_available():
# device = torch.device('cuda')
device = torch.device('cuda:{}'.format(torch.cuda.current_device()))
else:
device = torch.device('cpu')
# Log which device was actually used
log.log_message('Device used: '+str(device))
# Create a log for logging the loss values
log_prefix = 'log_train_epochs'
log_loss = log_prefix+'_losses'
log.create_log(log_loss, 'epoch', 'batch', 'loss', 'batch_train_acc')
# Obtain the dataset and dataloaders
trainloader, projectloader, testloader, classes, num_channels = get_dataloaders(args)
# Create a convolutional network based on arguments and add 1x1 conv layer
features_net, add_on_layers = get_network(num_channels, args)
# Create a ProtoTree
tree = ProtoTree(num_classes=len(classes),
feature_net = features_net,
args = args,
add_on_layers = add_on_layers)
tree = tree.to(device=device)
# Determine which optimizer should be used to update the tree parameters
optimizer, params_to_freeze, params_to_train = get_optimizer(tree, args)
scheduler = torch.optim.lr_scheduler.MultiStepLR(optimizer=optimizer, milestones=args.milestones, gamma=args.gamma)
tree, epoch = init_tree(tree, optimizer, scheduler, device, args)
tree.save(f'{log.checkpoint_dir}/tree_init')
log.log_message("Max depth %s, so %s internal nodes and %s leaves"%(args.depth, tree.num_branches, tree.num_leaves))
analyse_output_shape(tree, trainloader, log, device)
leaf_labels = dict()
best_train_acc = 0.
best_test_acc = 0.
if epoch < args.epochs + 1:
'''
TRAIN AND EVALUATE TREE
'''
for epoch in range(epoch, args.epochs + 1):
log.log_message("\nEpoch %s"%str(epoch))
# Freeze (part of) network for some epochs if indicated in args
freeze(tree, epoch, params_to_freeze, params_to_train, args, log)
log_learning_rates(optimizer, args, log)
# Train tree
if tree._kontschieder_train:
train_info = train_epoch_kontschieder(tree, trainloader, optimizer, epoch, args.disable_derivative_free_leaf_optim, device, log, log_prefix)
else:
train_info = train_epoch(tree, trainloader, optimizer, epoch, args.disable_derivative_free_leaf_optim, device, log, log_prefix)
save_tree(tree, optimizer, scheduler, epoch, log, args)
best_train_acc = save_best_train_tree(tree, optimizer, scheduler, best_train_acc, train_info['train_accuracy'], log)
leaf_labels = analyse_leafs(tree, epoch, len(classes), leaf_labels, args.pruning_threshold_leaves, log)
# Evaluate tree
if args.epochs>100:
if epoch%10==0 or epoch==args.epochs:
eval_info = eval(tree, testloader, epoch, device, log)
original_test_acc = eval_info['test_accuracy']
best_test_acc = save_best_test_tree(tree, optimizer, scheduler, best_test_acc, eval_info['test_accuracy'], log)
log.log_values('log_epoch_overview', epoch, eval_info['test_accuracy'], train_info['train_accuracy'], train_info['loss'])
else:
log.log_values('log_epoch_overview', epoch, "n.a.", train_info['train_accuracy'], train_info['loss'])
else:
eval_info = eval(tree, testloader, epoch, device, log)
original_test_acc = eval_info['test_accuracy']
best_test_acc = save_best_test_tree(tree, optimizer, scheduler, best_test_acc, eval_info['test_accuracy'], log)
log.log_values('log_epoch_overview', epoch, eval_info['test_accuracy'], train_info['train_accuracy'], train_info['loss'])
scheduler.step()
else: #tree was loaded and not trained, so evaluate only
'''
EVALUATE TREE
'''
eval_info = eval(tree, testloader, epoch, device, log)
original_test_acc = eval_info['test_accuracy']
best_test_acc = save_best_test_tree(tree, optimizer, scheduler, best_test_acc, eval_info['test_accuracy'], log)
log.log_values('log_epoch_overview', epoch, eval_info['test_accuracy'], "n.a.", "n.a.")
'''
EVALUATE AND ANALYSE TRAINED TREE
'''
log.log_message("Training Finished. Best training accuracy was %s, best test accuracy was %s\n"%(str(best_train_acc), str(best_test_acc)))
trained_tree = deepcopy(tree)
leaf_labels = analyse_leafs(tree, epoch+1, len(classes), leaf_labels, args.pruning_threshold_leaves, log)
analyse_leaf_distributions(tree, log)
'''
PRUNE
'''
pruned = prune(tree, args.pruning_threshold_leaves, log)
name = "pruned"
save_tree_description(tree, optimizer, scheduler, name, log)
pruned_tree = deepcopy(tree)
# Analyse and evaluate pruned tree
leaf_labels = analyse_leafs(tree, epoch+2, len(classes), leaf_labels, args.pruning_threshold_leaves, log)
analyse_leaf_distributions(tree, log)
eval_info = eval(tree, testloader, name, device, log)
pruned_test_acc = eval_info['test_accuracy']
pruned_tree = tree
'''
PROJECT
'''
project_info, tree = project_with_class_constraints(deepcopy(pruned_tree), projectloader, device, args, log)
name = "pruned_and_projected"
save_tree_description(tree, optimizer, scheduler, name, log)
pruned_projected_tree = deepcopy(tree)
# Analyse and evaluate pruned tree with projected prototypes
average_distance_nearest_image(project_info, tree, log)
leaf_labels = analyse_leafs(tree, epoch+3, len(classes), leaf_labels, args.pruning_threshold_leaves, log)
analyse_leaf_distributions(tree, log)
eval_info = eval(tree, testloader, name, device, log)
pruned_projected_test_acc = eval_info['test_accuracy']
eval_info_samplemax = eval(tree, testloader, name, device, log, 'sample_max')
get_avg_path_length(tree, eval_info_samplemax, log)
eval_info_greedy = eval(tree, testloader, name, device, log, 'greedy')
get_avg_path_length(tree, eval_info_greedy, log)
fidelity_info = eval_fidelity(tree, testloader, device, log)
# Upsample prototype for visualization
project_info = upsample(tree, project_info, projectloader, name, args, log)
# visualize tree
gen_vis(tree, name, args, classes)
return trained_tree.to('cpu'), pruned_tree.to('cpu'), pruned_projected_tree.to('cpu'), original_test_acc, pruned_test_acc, pruned_projected_test_acc, project_info, eval_info_samplemax, eval_info_greedy, fidelity_info
def train_epoch(tree: ProtoTree,
train_loader: DataLoader,
optimizer: torch.optim.Optimizer,
epoch: int,
disable_derivative_free_leaf_optim: bool,
device,
log: Log = None,
log_prefix: str = 'log_train_epochs',
progress_prefix: str = 'Train Epoch'
) -> dict:
tree = tree.to(device)
# Make sure the model is in eval mode
tree.eval()
# Store info about the procedure
train_info = dict()
total_loss = 0.
total_acc = 0.
# Create a log if required
log_loss = f'{log_prefix}_losses'
nr_batches = float(len(train_loader))
with torch.no_grad():
_old_dist_params = dict()
for leaf in tree.leaves:
_old_dist_params[leaf] = leaf._dist_params.detach().clone()
# Optimize class distributions in leafs
eye = torch.eye(tree._num_classes).to(device)
# Show progress on progress bar
train_iter = tqdm(enumerate(train_loader),
total=len(train_loader),
desc=progress_prefix+' %s'%epoch,
ncols=0)
# Iterate through the data set to update leaves, prototypes and network
for i, (xs, ys) in train_iter:
# Make sure the model is in train mode
tree.train()
# Reset the gradients
optimizer.zero_grad()
xs, ys = xs.to(device), ys.to(device)
# Perform a forward pass through the network
ys_pred, info = tree.forward(xs)
# Learn prototypes and network with gradient descent.
# If disable_derivative_free_leaf_optim, leaves are optimized with gradient descent as well.
# Compute the loss
if tree._log_probabilities:
loss = F.nll_loss(ys_pred, ys)
else:
loss = F.nll_loss(torch.log(ys_pred), ys)
# Compute the gradient
loss.backward()
# Update model parameters
optimizer.step()
if not disable_derivative_free_leaf_optim:
#Update leaves with derivate-free algorithm
#Make sure the tree is in eval mode
tree.eval()
with torch.no_grad():
target = eye[ys] #shape (batchsize, num_classes)
for leaf in tree.leaves:
if tree._log_probabilities:
# log version
update = torch.exp(torch.logsumexp(info['pa_tensor'][leaf.index] + leaf.distribution() + torch.log(target) - ys_pred, dim=0))
else:
update = torch.sum((info['pa_tensor'][leaf.index] * leaf.distribution() * target)/ys_pred, dim=0)
leaf._dist_params -= (_old_dist_params[leaf]/nr_batches)
F.relu_(leaf._dist_params) #dist_params values can get slightly negative because of floating point issues. therefore, set to zero.
leaf._dist_params += update
# Count the number of correct classifications
ys_pred_max = torch.argmax(ys_pred, dim=1)
correct = torch.sum(torch.eq(ys_pred_max, ys))
acc = correct.item() / float(len(xs))
train_iter.set_postfix_str(
f'Batch [{i + 1}/{len(train_loader)}], Loss: {loss.item():.3f}, Acc: {acc:.3f}'
)
# Compute metrics over this batch
total_loss+=loss.item()
total_acc+=acc
if log is not None:
log.log_values(log_loss, epoch, i + 1, loss.item(), acc)
train_info['loss'] = total_loss/float(i+1)
train_info['train_accuracy'] = total_acc/float(i+1)
return train_info
def train_epoch_kontschieder(tree: ProtoTree,
train_loader: DataLoader,
optimizer: torch.optim.Optimizer,
epoch: int,
disable_derivative_free_leaf_optim: bool,
device,
log: Log = None,
log_prefix: str = 'log_train_epochs',
progress_prefix: str = 'Train Epoch'
) -> dict:
tree = tree.to(device)
# Store info about the procedure
train_info = dict()
total_loss = 0.
total_acc = 0.
# Create a log if required
log_loss = f'{log_prefix}_losses'
if log is not None and epoch==1:
log.create_log(log_loss, 'epoch', 'batch', 'loss', 'batch_train_acc')
# Reset the gradients
optimizer.zero_grad()
if disable_derivative_free_leaf_optim:
print("WARNING: kontschieder arguments will be ignored when training leaves with gradient descent")
else:
if tree._kontschieder_normalization:
# Iterate over the dataset multiple times to learn leaves following Kontschieder's approach
for _ in range(10):
# Train leaves with derivative-free algorithm using normalization factor
train_leaves_epoch(tree, train_loader, epoch, device)
else:
# Train leaves with Kontschieder's derivative-free algorithm, but using softmax
train_leaves_epoch(tree, train_loader, epoch, device)
# Train prototypes and network.
# If disable_derivative_free_leaf_optim, leafs are optimized with gradient descent as well.
# Show progress on progress bar
train_iter = tqdm(enumerate(train_loader),
total=len(train_loader),
desc=progress_prefix+' %s'%epoch,
ncols=0)
# Make sure the model is in train mode
tree.train()
for i, (xs, ys) in train_iter:
xs, ys = xs.to(device), ys.to(device)
# Reset the gradients
optimizer.zero_grad()
# Perform a forward pass through the network
ys_pred, _ = tree.forward(xs)
# Compute the loss
if tree._log_probabilities:
loss = F.nll_loss(ys_pred, ys)
else:
loss = F.nll_loss(torch.log(ys_pred), ys)
# Compute the gradient
loss.backward()
# Update model parameters
optimizer.step()
# Count the number of correct classifications
ys_pred = torch.argmax(ys_pred, dim=1)
correct = torch.sum(torch.eq(ys_pred, ys))
acc = correct.item() / float(len(xs))
train_iter.set_postfix_str(
f'Batch [{i + 1}/{len(train_loader)}], Loss: {loss.item():.3f}, Acc: {acc:.3f}'
)
# Compute metrics over this batch
total_loss+=loss.item()
total_acc+=acc
if log is not None:
log.log_values(log_loss, epoch, i + 1, loss.item(), acc)
train_info['loss'] = total_loss/float(i+1)
train_info['train_accuracy'] = total_acc/float(i+1)
return train_info