0. 引言

在Braindecode系列中，我会介绍跟BCI IV 2a有关的所有相关示例。
在前面的章节中，我们介绍了Braindecode中，为训练模型创建了两种受支持的配置： trialwise decoding 和 cropped decoding。在本章节中，我会展示如何使用数据增强来训练EEG深度模型。它遵循了 trialwise decoding 示例，并且还说明了transform 对输入信号的影响。文章主要包括以下几个部分：

1. 加载和预处理数据集

   • 加载
   • 预处理
   • 提取窗口
   • 将数据集拆分为训练数据集和验证数据集

2. 定义一个Transform

   • 操作一个会话并可视化转换后的数据
   • 使用数据增强训练模型

3. 创建模型

   • 创建具有所需增强的脑电图调节器
   • 手动合成变换
   • 在数据集级别设置数据增强

注意：配置环境部分内容参考该系列的上一篇文章！！！

1. 加载和预处理数据集

1.1 加载

对数据集进行加载，具体代码如下：

from skorch.helper import predefined_split
from skorch.callbacks import LRScheduler
from braindecode import EEGClassifier
from braindecode.datasets import MOABBDataset
subject_id = 3
dataset = MOABBDataset(dataset_name="BNCI2014001", subject_ids=[subject_id])

1.2 预处理

对加载后的数据进行预处理操作，具体代码如下：

from braindecode.preprocessing import (
    exponential_moving_standardize, preprocess, Preprocessor)
from numpy import multiply
low_cut_hz = 4.  # low cut frequency for filtering
high_cut_hz = 38.  # high cut frequency for filtering
# Parameters for exponential moving standardization
factor_new = 1e-3
init_block_size = 1000
# Factor to convert from V to uV
factor = 1e6
preprocessors = [
    Preprocessor('pick_types', eeg=True, meg=False, stim=False),  # Keep EEG sensors
    Preprocessor(lambda data: multiply(data, factor)),  # Convert from V to uV
    Preprocessor('filter', l_freq=low_cut_hz, h_freq=high_cut_hz),  # Bandpass filter
    Preprocessor(exponential_moving_standardize,  # Exponential moving standardization
                 factor_new=factor_new, init_block_size=init_block_size)
]
preprocess(dataset, preprocessors)

1.3 提取窗口

对数据进行提取窗口操作，同trialwise decoding 示例，具体代码如下：

from braindecode.preprocessing import create_windows_from_events
trial_start_offset_seconds = -0.5
# Extract sampling frequency, check that they are same in all datasets
sfreq = dataset.datasets[0].raw.info['sfreq']
assert all([ds.raw.info['sfreq'] == sfreq for ds in dataset.datasets])
# Calculate the trial start offset in samples.
trial_start_offset_samples = int(trial_start_offset_seconds * sfreq)
# Create windows using braindecode function for this. It needs parameters to
# define how trials should be used.
windows_dataset = create_windows_from_events(
    dataset,
    trial_start_offset_samples=trial_start_offset_samples,
    trial_stop_offset_samples=0,
    preload=True,
)

1.4 将数据集拆分为训练数据集和验证数据集

将处理后的数据根据session进行分割，并将分割后的数据切分为训练集和验证集，具体代码如下：

splitted = windows_dataset.split('session')
train_set = splitted['session_T']
valid_set = splitted['session_E']

2.1 操作一个会话并可视化转换后的数据

接下来，让我们扩充一个会话，以显示由此产生的频率偏移。一个mne Epoch的数据用在这里来说明 mne function 的用法。

import torch
epochs = train_set.datasets[0].windows  # original epochs
X = epochs.get_data()
# This allows to apply the transform with a fixed shift (10 Hz) for
# visualization instead of sampling the shift randomly between -2 and 2 Hz
X_tr, _ = transform.operation(torch.as_tensor(X).float(), None, 10., sfreq)

转换后的会话的psd现在已经偏移了10 Hz，正如可以在psd图上看到的那样。具体代码如下：

import mne
import matplotlib.pyplot as plt
import numpy as np
def plot_psd(data, axis, label, color):
    psds, freqs = mne.time_frequency.psd_array_multitaper(data, sfreq=sfreq,
                                                          fmin=0.1, fmax=100)
    psds = 10. * np.log10(psds)
    psds_mean = psds.mean(0).mean(0)
    axis.plot(freqs, psds_mean, color=color, label=label)
_, ax = plt.subplots()
# 绘制原始psd图
plot_psd(X, ax, 'original', 'k')
# 绘制偏移后的psd图
plot_psd(X_tr.numpy(), ax, 'shifted', 'r')
ax.set(title='Multitaper PSD (gradiometers)', xlabel='Frequency (Hz)',
       ylabel='Power Spectral Density (dB)')
ax.legend()
plt.show()

生成的psd图如下：

2.2 使用数据增强训练模型

既然我们知道了如何实例化Transforms，现在是时候学习如何使用它们来训练模型并尝试提高其泛化能力了。具体内容在第三部分进行展示。

3. 创建模型

创建一个要训练的模型。具体代码如下：

from braindecode.util import set_random_seeds
from braindecode.models import ShallowFBCSPNet
cuda = torch.cuda.is_available()  # check if GPU is available, if True chooses to use it
device = 'cuda' if cuda else 'cpu'
if cuda:
    torch.backends.cudnn.benchmark = True
# Set random seed to be able to roughly reproduce results
# Note that with cudnn benchmark set to True, GPU indeterminism
# may still make results substantially different between runs.
# To obtain more consistent results at the cost of increased computation time,
# you can set `cudnn_benchmark=False` in `set_random_seeds`
# or remove `torch.backends.cudnn.benchmark = True`
seed = 20200220
set_random_seeds(seed=seed, cuda=cuda)
n_classes = 4
# Extract number of chans and time steps from dataset
n_channels = train_set[0][0].shape[0]
input_window_samples = train_set[0][0].shape[1]
model = ShallowFBCSPNet(
    n_channels,
    n_classes,
    input_window_samples=input_window_samples,
    final_conv_length='auto',
)

3.1 创建具有所需增强的脑电图调节器

为了使用数据扩充进行训练，可以使用自定义数据加载器进行训练。多个Transforms可以传递给它，并将依次应用于AugmentedDataLoader对象中的批处理数据。
首先，定义模型中使用的transforms 的功能结构。具体代码如下：

from braindecode.augmentation import AugmentedDataLoader, SignFlip
freq_shift = FrequencyShift(
    probability=.5,
    sfreq=sfreq,
    max_delta_freq=2.  # the frequency shifts are sampled now between -2 and 2 Hz
)
sign_flip = SignFlip(probability=.1)
transforms = [
    freq_shift,
    sign_flip
]

# Send model to GPU
if cuda:
model.cuda()

然后，将该transforms用于模型的训练之中。该模型训练实例遵循trial-wise示例。与之不同的是，AugmentedDataLoader用作train iterator，transforms 列表作为参数传递。具体代码如下：
注意：这里AugmentedDataLoader的意思是为了说明输入数据进行了数据增强，并没有明确的定义！！！

lr = 0.0625 * 0.01
weight_decay = 0
batch_size = 64
n_epochs = 4
clf = EEGClassifier(
    model,
    iterator_train=AugmentedDataLoader,  # This tells EEGClassifier to use a custom DataLoader
    iterator_train__transforms=transforms,  # This sets the augmentations to use
    criterion=torch.nn.NLLLoss,
    optimizer=torch.optim.AdamW,
    train_split=predefined_split(valid_set),  # using valid_set for validation
    optimizer__lr=lr,
    optimizer__weight_decay=weight_decay,
    batch_size=batch_size,
    callbacks=[
        "accuracy",
        ("lr_scheduler", LRScheduler('CosineAnnealingLR', T_max=n_epochs - 1)),
    ],
    device=device,
)
# Model training for a specified number of epochs. `y` is None as it is already
# supplied in the dataset.
clf.fit(train_set, y=None, epochs=n_epochs)

3.2 手动合成变换

上述变换同组合传递是等效的，即：3.1节 = 3.2节+3.3节的内容。将相同转换的组合传递给EEG分类器是等效的（尽管更详细）：

from braindecode.augmentation import Compose

composed_transforms = Compose(transforms=transforms)