minerva.models.nets.cpc_networks

Classes

CNN	Base class for all neural network modules.
ConvBlock	Base class for all neural network modules.
Convolutional1DEncoder	Base class for all neural network modules.
Genc_Gar	Base class for all neural network modules.
HARCPCAutoregressive	Base class for all neural network modules.
HARPredictionHead	Base class for all neural network modules.
LinearClassifier	Base class for all neural network modules.
PredictionNetwork	Base class for all neural network modules.
ResNetEncoder	Base class for all neural network modules.

Module Contents

class minerva.models.nets.cpc_networks.CNN

Bases: lightning.LightningModule

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing them to be nested in a tree structure. You can assign the submodules as regular attributes:

import torch.nn as nn
import torch.nn.functional as F

class Model(nn.Module):
    def __init__(self) -> None:
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will also have their parameters converted when you call to(), etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

Variables:

training (bool) – Boolean represents whether this module is in training or evaluation mode.

Convolutional Neural Network (CNN) encoder for CPC (Contrastive Predictive Coding) for Human Activity Recognition (HAR).

This class serves as a wrapper for the Convolutional1DEncoder class, providing an easy-to-use interface for the CPC model.

encoder

forward(x)

Same as torch.nn.Module.forward().

Args:

*args: Whatever you decide to pass into the forward method. **kwargs: Keyword arguments are also possible.

Return:

Your model’s output

class minerva.models.nets.cpc_networks.ConvBlock(in_channels=6, out_channels=128, kernel_size=1, stride=1, padding=1, padding_mode='reflect', dropout_prob=0.2)

Bases: lightning.LightningModule

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing them to be nested in a tree structure. You can assign the submodules as regular attributes:

import torch.nn as nn
import torch.nn.functional as F

class Model(nn.Module):
    def __init__(self) -> None:
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will also have their parameters converted when you call to(), etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

Variables:

training (bool) – Boolean represents whether this module is in training or evaluation mode.

Convolutional Block for the 1D Convolutional Encoder.

This block consists of a convolutional layer followed by a ReLU activation and dropout.

Parameters

in_channelsint, optional

Number of input channels, by default 6.

out_channelsint, optional

Number of output channels, by default 128.

kernel_sizeint, optional

Size of the convolutional kernel, by default 1.

strideint, optional

Stride of the convolution, by default 1.

paddingint, optional

Padding for the convolution, by default 1.

padding_modestr, optional

Padding mode for the convolution, by default ‘reflect’.

dropout_probfloat, optional

Dropout probability, by default 0.2.

conv

dropout

forward(inputs)

Same as torch.nn.Module.forward().

Args:

*args: Whatever you decide to pass into the forward method. **kwargs: Keyword arguments are also possible.

Return:

Your model’s output

relu

class minerva.models.nets.cpc_networks.Convolutional1DEncoder(input_size=6, kernel_size=3, stride=1, padding=1)

Bases: lightning.LightningModule

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing them to be nested in a tree structure. You can assign the submodules as regular attributes:

import torch.nn as nn
import torch.nn.functional as F

class Model(nn.Module):
    def __init__(self) -> None:
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will also have their parameters converted when you call to(), etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

Variables:

training (bool) – Boolean represents whether this module is in training or evaluation mode.

1D Convolutional Encoder for CPC.

This encoder consists of a sequence of convolutional blocks that process the input time series data.

Parameters

input_sizeint, optional

Number of input channels, by default 6.

kernel_sizeint, optional

Size of the convolutional kernel, by default 3.

strideint, optional

Stride of the convolution, by default 1.

paddingint, optional

Padding for the convolution, by default 1.

encoder

forward(x)

Same as torch.nn.Module.forward().

Args:

*args: Whatever you decide to pass into the forward method. **kwargs: Keyword arguments are also possible.

Return:

Your model’s output

class minerva.models.nets.cpc_networks.Genc_Gar(g_enc, g_ar)

Bases: torch.nn.Module

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing them to be nested in a tree structure. You can assign the submodules as regular attributes:

import torch.nn as nn
import torch.nn.functional as F

class Model(nn.Module):
    def __init__(self) -> None:
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will also have their parameters converted when you call to(), etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

Variables:

training (bool) – Boolean represents whether this module is in training or evaluation mode.

Parameters:

• g_enc (torch.nn.Module)

• g_ar (torch.nn.Module)

Combination of the GENC (encoder) and GAR (autoregressive) networks, forming the backbone of the CPC model for HAR.

Parameters

g_enc: torch.nn.Module

Encoder network to extract features from the input data.

g_artorch.nn.Module

Autoregressive network to model temporal dependencies in the feature space.

forward(x)

g_ar

g_enc

class minerva.models.nets.cpc_networks.HARCPCAutoregressive(input_size=128, hidden_size=256, num_layers=2, bidirectional=False, batch_first=True, dropout=0.2)

Bases: lightning.LightningModule

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing them to be nested in a tree structure. You can assign the submodules as regular attributes:

import torch.nn as nn
import torch.nn.functional as F

class Model(nn.Module):
    def __init__(self) -> None:
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will also have their parameters converted when you call to(), etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

Variables:

training (bool) – Boolean represents whether this module is in training or evaluation mode.

Autoregressive model for CPC used in Human Activity Recognition (HAR).

This network models the temporal dependencies in the feature space.

Parameters

input_sizeint, optional

Number of input features, by default 128.

hidden_sizeint, optional

Number of hidden units, by default 256.

num_layersint, optional

Number of recurrent layers, by default 2.

bidirectionalbool, optional

If True, becomes a bidirectional GRU, by default False.

batch_firstbool, optional

If True, the input and output tensors are provided as (batch, seq, feature), by default True.

dropoutfloat, optional

Dropout probability, by default 0.2.

forward(x, hidden=None)

Same as torch.nn.Module.forward().

Args:

*args: Whatever you decide to pass into the forward method. **kwargs: Keyword arguments are also possible.

Return:

Your model’s output

rnn

class minerva.models.nets.cpc_networks.HARPredictionHead(num_classes=9)

Bases: lightning.LightningModule

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing them to be nested in a tree structure. You can assign the submodules as regular attributes:

import torch.nn as nn
import torch.nn.functional as F

class Model(nn.Module):
    def __init__(self) -> None:
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will also have their parameters converted when you call to(), etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

Variables:

training (bool) – Boolean represents whether this module is in training or evaluation mode.

Parameters:

num_classes (int)

Prediction head for Human Activity Recognition (HAR).

This network takes the encoded and temporally modeled features and outputs the final activity classification.

Parameters

num_classesint, optional

Number of activity classes to predict, by default 9 (RW_waist).

forward(x)

Same as torch.nn.Module.forward().

Args:

*args: Whatever you decide to pass into the forward method. **kwargs: Keyword arguments are also possible.

Return:

Your model’s output

model

class minerva.models.nets.cpc_networks.LinearClassifier(backbone, head, num_classes=6, learning_rate=0.001, flatten=True, freeze_backbone=False, loss_fn=None)

Bases: lightning.LightningModule

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing them to be nested in a tree structure. You can assign the submodules as regular attributes:

import torch.nn as nn
import torch.nn.functional as F

class Model(nn.Module):
    def __init__(self) -> None:
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will also have their parameters converted when you call to(), etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

Variables:

training (bool) – Boolean represents whether this module is in training or evaluation mode.

Parameters:

• backbone (lightning.LightningModule)

• head (lightning.LightningModule)

• num_classes (int)

• learning_rate (float)

• flatten (bool)

• freeze_backbone (bool)

• loss_fn (torch.nn.modules.loss._Loss)

A linear classifier model built on top of a backbone and a head network, designed for tasks such as classification. This model leverages PyTorch Lightning for easier training and evaluation.

Parameters

backboneL.LightningModule

The backbone network used for feature extraction.

headL.LightningModule

The head network used for the final classification.

num_classesint, optional

The number of output classes, by default 6.

learning_ratefloat, optional

The learning rate for the optimizer, by default 0.001.

flattenbool, optional

Whether to flatten the output of the backbone before passing it to the head, by default True.

freeze_backbonebool, optional

Whether to freeze the backbone during training, by default False.

loss_fntorch.nn.modules.loss._Loss, optional

The loss function to use, by default CrossEntropyLoss.

_freeze(model)

Freezes the model, i.e. sets the requires_grad parameter of all the parameters to False.

Parameters

modeltype

The model to freeze

backbone

calculate_metrics(y_pred, y_true, stage_name)

Calculate metrics for the given batch.

Parameters

y_predtorch.Tensor

Predicted labels.

y_truetorch.Tensor

True labels.

Returns

dict

Dictionary of metrics.

Parameters:

• y_pred (torch.Tensor)

• y_true (torch.Tensor)

• stage_name (str)

Return type:

dict

configure_optimizers()

Configures the optimizer. If update_backbone is True, it will update the parameters of the backbone and the head. Otherwise, it will only update the parameters of the head.

flatten = True

forward(x)

Same as torch.nn.Module.forward().

Args:

*args: Whatever you decide to pass into the forward method. **kwargs: Keyword arguments are also possible.

Return:

Your model’s output

Parameters:

x (torch.Tensor)

freeze_backbone = False

head

learning_rate = 0.001

loss_fn = None

num_classes = 6

test_step(batch, batch_idx)

Operates on a single batch of data from the test set. In this step you’d normally generate examples or calculate anything of interest such as accuracy.

Args:

batch: The output of your data iterable, normally a DataLoader. batch_idx: The index of this batch. dataloader_idx: The index of the dataloader that produced this batch.

(only if multiple dataloaders used)

Return:

• Tensor - The loss tensor

• dict - A dictionary. Can include any keys, but must include the key 'loss'.

• None - Skip to the next batch.

# if you have one test dataloader:
def test_step(self, batch, batch_idx): ...


# if you have multiple test dataloaders:
def test_step(self, batch, batch_idx, dataloader_idx=0): ...

Examples:

# CASE 1: A single test dataset
def test_step(self, batch, batch_idx):
    x, y = batch

    # implement your own
    out = self(x)
    loss = self.loss(out, y)

    # log 6 example images
    # or generated text... or whatever
    sample_imgs = x[:6]
    grid = torchvision.utils.make_grid(sample_imgs)
    self.logger.experiment.add_image('example_images', grid, 0)

    # calculate acc
    labels_hat = torch.argmax(out, dim=1)
    test_acc = torch.sum(y == labels_hat).item() / (len(y) * 1.0)

    # log the outputs!
    self.log_dict({'test_loss': loss, 'test_acc': test_acc})

If you pass in multiple test dataloaders, test_step() will have an additional argument. We recommend setting the default value of 0 so that you can quickly switch between single and multiple dataloaders.

# CASE 2: multiple test dataloaders
def test_step(self, batch, batch_idx, dataloader_idx=0):
    # dataloader_idx tells you which dataset this is.
    ...

Note:

If you don’t need to test you don’t need to implement this method.

Note:

When the test_step() is called, the model has been put in eval mode and PyTorch gradients have been disabled. At the end of the test epoch, the model goes back to training mode and gradients are enabled.

Parameters:

• batch (torch.Tensor)

• batch_idx (int)

training_step(batch, batch_idx)

Here you compute and return the training loss and some additional metrics for e.g. the progress bar or logger.

Args:

batch: The output of your data iterable, normally a DataLoader. batch_idx: The index of this batch. dataloader_idx: The index of the dataloader that produced this batch.

(only if multiple dataloaders used)

Return:

• Tensor - The loss tensor

• dict - A dictionary which can include any keys, but must include the key 'loss' in the case of automatic optimization.

• None - In automatic optimization, this will skip to the next batch (but is not supported for multi-GPU, TPU, or DeepSpeed). For manual optimization, this has no special meaning, as returning the loss is not required.

In this step you’d normally do the forward pass and calculate the loss for a batch. You can also do fancier things like multiple forward passes or something model specific.

Example:

def training_step(self, batch, batch_idx):
    x, y, z = batch
    out = self.encoder(x)
    loss = self.loss(out, x)
    return loss

To use multiple optimizers, you can switch to ‘manual optimization’ and control their stepping:

def __init__(self):
    super().__init__()
    self.automatic_optimization = False


# Multiple optimizers (e.g.: GANs)
def training_step(self, batch, batch_idx):
    opt1, opt2 = self.optimizers()

    # do training_step with encoder
    ...
    opt1.step()
    # do training_step with decoder
    ...
    opt2.step()

Note:

When accumulate_grad_batches > 1, the loss returned here will be automatically normalized by accumulate_grad_batches internally.

Parameters:

• batch (torch.Tensor)

• batch_idx (int)

validation_step(batch, batch_idx)

Operates on a single batch of data from the validation set. In this step you’d might generate examples or calculate anything of interest like accuracy.

Args:

batch: The output of your data iterable, normally a DataLoader. batch_idx: The index of this batch. dataloader_idx: The index of the dataloader that produced this batch.

(only if multiple dataloaders used)

Return:

• Tensor - The loss tensor

• dict - A dictionary. Can include any keys, but must include the key 'loss'.

• None - Skip to the next batch.

# if you have one val dataloader:
def validation_step(self, batch, batch_idx): ...


# if you have multiple val dataloaders:
def validation_step(self, batch, batch_idx, dataloader_idx=0): ...

Examples:

# CASE 1: A single validation dataset
def validation_step(self, batch, batch_idx):
    x, y = batch

    # implement your own
    out = self(x)
    loss = self.loss(out, y)

    # log 6 example images
    # or generated text... or whatever
    sample_imgs = x[:6]
    grid = torchvision.utils.make_grid(sample_imgs)
    self.logger.experiment.add_image('example_images', grid, 0)

    # calculate acc
    labels_hat = torch.argmax(out, dim=1)
    val_acc = torch.sum(y == labels_hat).item() / (len(y) * 1.0)

    # log the outputs!
    self.log_dict({'val_loss': loss, 'val_acc': val_acc})

If you pass in multiple val dataloaders, validation_step() will have an additional argument. We recommend setting the default value of 0 so that you can quickly switch between single and multiple dataloaders.

# CASE 2: multiple validation dataloaders
def validation_step(self, batch, batch_idx, dataloader_idx=0):
    # dataloader_idx tells you which dataset this is.
    ...

Note:

If you don’t need to validate you don’t need to implement this method.

Note:

When the validation_step() is called, the model has been put in eval mode and PyTorch gradients have been disabled. At the end of validation, the model goes back to training mode and gradients are enabled.

Parameters:

• batch (torch.Tensor)

• batch_idx (int)

class minerva.models.nets.cpc_networks.PredictionNetwork(in_channels=256, out_channels=128)

Bases: lightning.LightningModule

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing them to be nested in a tree structure. You can assign the submodules as regular attributes:

import torch.nn as nn
import torch.nn.functional as F

class Model(nn.Module):
    def __init__(self) -> None:
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will also have their parameters converted when you call to(), etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

Variables:

training (bool) – Boolean represents whether this module is in training or evaluation mode.

Projection head for CPC used in Human Activity Recognition (HAR).

This network projects the encoded representations to a lower-dimensional space to facilitate the contrastive learning process.

Parameters

in_channelsint, optional

Number of input channels, by default 256.

out_channelsint, optional

Number of output channels, by default 128.

Wk

forward(x)

Same as torch.nn.Module.forward().

Args:

*args: Whatever you decide to pass into the forward method. **kwargs: Keyword arguments are also possible.

Return:

Your model’s output

class minerva.models.nets.cpc_networks.ResNetEncoder(permute=True, *args, **kwargs)

Bases: minerva.models.nets.time_series.resnet._ResNet1D

Base class for all neural network modules.

Your models should also subclass this class.

Modules can also contain other Modules, allowing them to be nested in a tree structure. You can assign the submodules as regular attributes:

import torch.nn as nn
import torch.nn.functional as F

class Model(nn.Module):
    def __init__(self) -> None:
        super().__init__()
        self.conv1 = nn.Conv2d(1, 20, 5)
        self.conv2 = nn.Conv2d(20, 20, 5)

    def forward(self, x):
        x = F.relu(self.conv1(x))
        return F.relu(self.conv2(x))

Submodules assigned in this way will be registered, and will also have their parameters converted when you call to(), etc.

Note

As per the example above, an __init__() call to the parent class must be made before assignment on the child.

Variables:

training (bool) – Boolean represents whether this module is in training or evaluation mode.

Initialize internal Module state, shared by both nn.Module and ScriptModule.

forward(x)

permute = True