From 70ad544238171fdcb54cf63f09ba55f7a0b2843c Mon Sep 17 00:00:00 2001
From: Joerg Martin <joerg.martin@ptb.de>
Date: Fri, 26 Nov 2021 15:47:39 +0100
Subject: [PATCH] NonEiV for energy effiency and concrete strength added
Both reaching similar RMSEs as in MC Dropout paper.
Updated hidden_layers handling in all training files.
---
Experiments/evaluate_energy.py | 45 ++++++
...lifornia.py => train_noneiv_california.py} | 3 +-
Experiments/train_noneiv_concrete.py | 147 ++++++++++++++++++
Experiments/train_noneiv_energy.py | 147 ++++++++++++++++++
4 files changed, 341 insertions(+), 1 deletion(-)
create mode 100644 Experiments/evaluate_energy.py
rename Experiments/{train_noneiv_carlifornia.py => train_noneiv_california.py} (97%)
create mode 100644 Experiments/train_noneiv_concrete.py
create mode 100644 Experiments/train_noneiv_energy.py
diff --git a/Experiments/evaluate_energy.py b/Experiments/evaluate_energy.py
new file mode 100644
index 0000000..f37b094
--- /dev/null
+++ b/Experiments/evaluate_energy.py
@@ -0,0 +1,45 @@
+import os
+import numpy as np
+import torch
+import torch.backends.cudnn
+from torch.utils.data import DataLoader
+from torch.utils.tensorboard.writer import SummaryWriter
+
+from EIVArchitectures import Networks, initialize_weights
+from EIVData.energy_efficiency import load_data
+from EIVTrainingRoutines import train_and_store, loss_functions
+
+from train_noneiv_energy import p, init_std_y_list, seed_list, unscaled_reg, hidden_layers
+
+
+train_data, test_data = load_data()
+test_dataloader = DataLoader(test_data, batch_size=int(np.max((len(test_data), 800))))
+
+seed = seed_list[0]
+init_std_y = init_std_y_list[0]
+saved_file = os.path.join('saved_networks',
+ f'noneiv_energy'\
+ f'init_std_y_{init_std_y:.3f}_ureg_{unscaled_reg:.1f}'\
+ f'_p_{p:.2f}_seed_{seed}.pkl')
+
+input_dim = train_data[0][0].numel()
+output_dim = train_data[0][1].numel()
+net = Networks.FNNBer(p=p, init_std_y=init_std_y,
+ h=[input_dim, *hidden_layers, output_dim])
+train_and_store.open_stored_training(saved_file=saved_file,
+ net=net)
+
+
+# RMSE
+x,y = next(iter(test_dataloader))
+out = net(x)[0]
+if len(y.shape) <=1:
+ y = y.view((-1,1))
+assert y.shape == out.shape
+res = y-out
+scale = train_data.dataset.std_labels
+scaled_res = res * scale.view((1,-1))
+scaled_res = scaled_res.detach().cpu().numpy().flatten()
+rmse = np.sqrt(np.mean(scaled_res**2))
+print(f'RMSE {rmse:.3f}')
+
diff --git a/Experiments/train_noneiv_carlifornia.py b/Experiments/train_noneiv_california.py
similarity index 97%
rename from Experiments/train_noneiv_carlifornia.py
rename to Experiments/train_noneiv_california.py
index 6ce2673..87ad6c4 100644
--- a/Experiments/train_noneiv_carlifornia.py
+++ b/Experiments/train_noneiv_california.py
@@ -27,6 +27,7 @@ lr_update = 20
epoch_offset = 10
init_std_y_list = [0.5]
gamma = 0.5
+hidden_layers = [1024, 1024, 1024, 1024]
device = torch.device('cuda:1' if torch.cuda.is_available() else 'cpu')
# reproducability
@@ -107,7 +108,7 @@ def train_on_data(init_std_y, seed):
output_dim = train_data[0][1].numel()
net = Networks.FNNBer(p=p,
init_std_y=init_std_y,
- h=[input_dim, 1024, 1024, 1024, 1024, output_dim])
+ h=[input_dim, *hidden_layers, output_dim])
net.apply(initialize_weights.glorot_init)
net = net.to(device)
net.std_y_par.requires_grad = False
diff --git a/Experiments/train_noneiv_concrete.py b/Experiments/train_noneiv_concrete.py
new file mode 100644
index 0000000..625db7d
--- /dev/null
+++ b/Experiments/train_noneiv_concrete.py
@@ -0,0 +1,147 @@
+"""
+Train non-EiV model on concrete strength dataset using different seeds
+"""
+import random
+import os
+
+import numpy as np
+import torch
+import torch.backends.cudnn
+from torch.utils.data import DataLoader
+from torch.utils.tensorboard.writer import SummaryWriter
+
+from EIVArchitectures import Networks, initialize_weights
+from EIVData.concrete_strength import load_data
+from EIVTrainingRoutines import train_and_store, loss_functions
+
+# hyperparameters
+lr = 1e-3
+batch_size = 32
+test_batch_size = 800
+number_of_epochs = 100
+unscaled_reg = 10
+report_point = 5
+p = 0.2
+lr_update = 20
+# pretraining = 300
+epoch_offset = 10
+init_std_y_list = [0.5]
+gamma = 0.5
+hidden_layers = [1024, 1024, 1024, 1024]
+device = torch.device('cuda:1' if torch.cuda.is_available() else 'cpu')
+
+# reproducability
+def set_seeds(seed):
+ torch.backends.cudnn.benchmark = False
+ np.random.seed(seed)
+ random.seed(seed)
+ torch.manual_seed(seed)
+seed_list = [0,]
+
+# to store the RMSE
+rmse_chain = []
+
+class UpdatedTrainEpoch(train_and_store.TrainEpoch):
+ def pre_epoch_update(self, net, epoch):
+ """
+ Overwrites the corresponding method
+ """
+ if epoch == 0:
+ self.lr = self.initial_lr
+ self.optimizer = torch.optim.Adam(net.parameters(), lr=self.lr)
+ self.lr_scheduler = torch.optim.lr_scheduler.StepLR(
+ self.optimizer, lr_update, gamma)
+
+
+ def post_epoch_update(self, net, epoch):
+ """
+ Overwrites the corresponding method
+ """
+ if epoch >= epoch_offset:
+ net.std_y_par.requires_grad = True
+ self.lr_scheduler.step()
+
+ def extra_report(self, net, i):
+ """
+ Overwrites the corresponding method
+ and fed after initialization of this class
+ """
+ rmse = self.rmse(net).item()
+ rmse_chain.append(rmse)
+ writer.add_scalar('RMSE', rmse, self.total_count)
+ writer.add_scalar('train loss', self.last_train_loss, self.total_count)
+ writer.add_scalar('test loss', self.last_test_loss, self.total_count)
+ print(f'RMSE {rmse:.3f}')
+
+ def rmse(self, net):
+ """
+ Compute the root mean squared error for `net`
+ """
+ net_train_state = net.training
+ net.eval()
+ x, y = next(iter(self.test_dataloader))
+ if len(y.shape) <= 1:
+ y = y.view((-1,1))
+ out = net(x.to(device))[0].detach().cpu()
+ assert out.shape == y.shape
+ if net_train_state:
+ net.train()
+ return torch.sqrt(torch.mean((out-y)**2))
+
+def train_on_data(init_std_y, seed):
+ """
+ Sets `seed`, loads data and trains an Bernoulli Modell, starting with
+ `init_std_y`.
+ """
+ # set seed
+ set_seeds(seed)
+ # load Datasets
+ train_data, test_data = load_data(seed=seed, splitting_part=0.8,
+ normalize=True)
+ # make dataloaders
+ train_dataloader = DataLoader(train_data, batch_size=batch_size,
+ shuffle=True)
+ test_dataloader = DataLoader(test_data, batch_size=test_batch_size,
+ shuffle=True)
+ # create a net
+ input_dim = train_data[0][0].numel()
+ output_dim = train_data[0][1].numel()
+ net = Networks.FNNBer(p=p,
+ init_std_y=init_std_y,
+ h=[input_dim, *hidden_layers, output_dim])
+ net.apply(initialize_weights.glorot_init)
+ net = net.to(device)
+ net.std_y_par.requires_grad = False
+ std_x_map = lambda: 0.0
+ std_y_map = lambda: net.get_std_y().detach().cpu().item()
+ # regularization
+ reg = unscaled_reg/len(train_data)
+ # create epoch_map
+ criterion = loss_functions.nll_reg_loss
+ epoch_map = UpdatedTrainEpoch(train_dataloader=train_dataloader,
+ test_dataloader=test_dataloader,
+ criterion=criterion, std_y_map=std_y_map, std_x_map=std_x_map,
+ lr=lr, reg=reg, report_point=report_point, device=device)
+ # run and save
+ save_file = os.path.join('saved_networks',
+ f'noneiv_concrete'\
+ f'init_std_y_{init_std_y:.3f}_ureg_{unscaled_reg:.1f}'\
+ f'_p_{p:.2f}_seed_{seed}.pkl')
+ train_and_store.train_and_store(net=net,
+ epoch_map=epoch_map,
+ number_of_epochs=number_of_epochs,
+ save_file=save_file)
+
+
+if __name__ == '__main__':
+ for seed in seed_list:
+ # Tensorboard monitoring
+ writer = SummaryWriter(log_dir=f'/home/martin09/tmp/tensorboard/'\
+ f'run_noneiv_concrete_lr_{lr:.4f}_seed'\
+ f'_{seed}_uregu_{unscaled_reg:.1f}_p_{p:.2f}')
+ print(f'>>>>SEED: {seed}')
+ for init_std_y in init_std_y_list:
+ print(f'Using init_std_y={init_std_y:.3f}')
+ train_on_data(init_std_y, seed)
+
+
diff --git a/Experiments/train_noneiv_energy.py b/Experiments/train_noneiv_energy.py
new file mode 100644
index 0000000..a635741
--- /dev/null
+++ b/Experiments/train_noneiv_energy.py
@@ -0,0 +1,147 @@
+"""
+Train non-EiV model on energy efficiency dataset using different seeds
+"""
+import random
+import os
+
+import numpy as np
+import torch
+import torch.backends.cudnn
+from torch.utils.data import DataLoader
+from torch.utils.tensorboard.writer import SummaryWriter
+
+from EIVArchitectures import Networks, initialize_weights
+from EIVData.energy_efficiency import load_data
+from EIVTrainingRoutines import train_and_store, loss_functions
+
+# hyperparameters
+lr = 1e-3
+batch_size = 32
+test_batch_size = 600
+number_of_epochs = 300
+unscaled_reg = 10
+report_point = 5
+p = 0.2
+lr_update = 50
+# pretraining = 300
+epoch_offset = 50
+init_std_y_list = [0.5]
+gamma = 0.5
+hidden_layers = [1024, 1024, 1024, 1024]
+device = torch.device('cuda:1' if torch.cuda.is_available() else 'cpu')
+
+# reproducability
+def set_seeds(seed):
+ torch.backends.cudnn.benchmark = False
+ np.random.seed(seed)
+ random.seed(seed)
+ torch.manual_seed(seed)
+seed_list = [0,]
+
+# to store the RMSE
+rmse_chain = []
+
+class UpdatedTrainEpoch(train_and_store.TrainEpoch):
+ def pre_epoch_update(self, net, epoch):
+ """
+ Overwrites the corresponding method
+ """
+ if epoch == 0:
+ self.lr = self.initial_lr
+ self.optimizer = torch.optim.Adam(net.parameters(), lr=self.lr)
+ self.lr_scheduler = torch.optim.lr_scheduler.StepLR(
+ self.optimizer, lr_update, gamma)
+
+
+ def post_epoch_update(self, net, epoch):
+ """
+ Overwrites the corresponding method
+ """
+ if epoch >= epoch_offset:
+ net.std_y_par.requires_grad = True
+ self.lr_scheduler.step()
+
+ def extra_report(self, net, i):
+ """
+ Overwrites the corresponding method
+ and fed after initialization of this class
+ """
+ rmse = self.rmse(net).item()
+ rmse_chain.append(rmse)
+ writer.add_scalar('RMSE', rmse, self.total_count)
+ writer.add_scalar('train loss', self.last_train_loss, self.total_count)
+ writer.add_scalar('test loss', self.last_test_loss, self.total_count)
+ print(f'RMSE {rmse:.3f}')
+
+ def rmse(self, net):
+ """
+ Compute the root mean squared error for `net`
+ """
+ net_train_state = net.training
+ net.eval()
+ x, y = next(iter(self.test_dataloader))
+ if len(y.shape) <= 1:
+ y = y.view((-1,1))
+ out = net(x.to(device))[0].detach().cpu()
+ assert out.shape == y.shape
+ if net_train_state:
+ net.train()
+ return torch.sqrt(torch.mean((out-y)**2))
+
+def train_on_data(init_std_y, seed):
+ """
+ Sets `seed`, loads data and trains an Bernoulli Modell, starting with
+ `init_std_y`.
+ """
+ # set seed
+ set_seeds(seed)
+ # load Datasets
+ train_data, test_data = load_data(seed=seed, splitting_part=0.8,
+ normalize=True)
+ # make dataloaders
+ train_dataloader = DataLoader(train_data, batch_size=batch_size,
+ shuffle=True)
+ test_dataloader = DataLoader(test_data, batch_size=test_batch_size,
+ shuffle=True)
+ # create a net
+ input_dim = train_data[0][0].numel()
+ output_dim = train_data[0][1].numel()
+ net = Networks.FNNBer(p=p,
+ init_std_y=init_std_y,
+ h=[input_dim, *hidden_layers, output_dim])
+ net.apply(initialize_weights.glorot_init)
+ net = net.to(device)
+ net.std_y_par.requires_grad = False
+ std_x_map = lambda: 0.0
+ std_y_map = lambda: net.get_std_y().detach().cpu().item()
+ # regularization
+ reg = unscaled_reg/len(train_data)
+ # create epoch_map
+ criterion = loss_functions.nll_reg_loss
+ epoch_map = UpdatedTrainEpoch(train_dataloader=train_dataloader,
+ test_dataloader=test_dataloader,
+ criterion=criterion, std_y_map=std_y_map, std_x_map=std_x_map,
+ lr=lr, reg=reg, report_point=report_point, device=device)
+ # run and save
+ save_file = os.path.join('saved_networks',
+ f'noneiv_energy'\
+ f'init_std_y_{init_std_y:.3f}_ureg_{unscaled_reg:.1f}'\
+ f'_p_{p:.2f}_seed_{seed}.pkl')
+ train_and_store.train_and_store(net=net,
+ epoch_map=epoch_map,
+ number_of_epochs=number_of_epochs,
+ save_file=save_file)
+
+
+if __name__ == '__main__':
+ for seed in seed_list:
+ # Tensorboard monitoring
+ writer = SummaryWriter(log_dir=f'/home/martin09/tmp/tensorboard/'\
+ f'run_noneiv_energy_lr_{lr:.4f}_seed'\
+ f'_{seed}_uregu_{unscaled_reg:.1f}_p_{p:.2f}')
+ print(f'>>>>SEED: {seed}')
+ for init_std_y in init_std_y_list:
+ print(f'Using init_std_y={init_std_y:.3f}')
+ train_on_data(init_std_y, seed)
+
+
--
GitLab