I am trying to implement classification of ECG segments from PTB-XL database (https://physionet.org/content/ptb-xl/1.0.3/). The architecture of the model which I am using is: import torch import torch.nn as nn class ResNetBlock(nn.Module): def __init__(self, in_channels, out_channels): super().__init__() self.conv1 = nn.Conv1d(in_channels, out_channels, kernel_size=7, padding=3,stride=1) self.bn1 = nn.BatchNorm1d(out_channels) self.conv2 = nn.Conv1d(out_channels, out_channels, kernel_size=5, padding=2,stride=1) self.bn2 = nn.BatchNorm1d(out_channels) self.conv3 = nn.Conv1d(in_channels=out_channels,out_channels=out_channels,kernel_size=3,padding=1,stride=1) self.bn3 = nn.BatchNorm1d(out_channels) skip_layers = [] if in_channels != out_channels: skip_layers += [ nn.Conv1d(in_channels, out_channels, kernel_size=1, stride=1), nn.BatchNorm1d(out_channels) ] self.skip_bn = nn.BatchNorm1d(out_channels) self.skip = nn.Sequential(*skip_layers) self.relu = nn.ReLU(inplace=True) class Model(nn.Module): def __init__(self,in_channels,num_classes): super().__init__() self.block_sizes = [64,128,256] self.resnet_block1 = ResNetBlock(in_channels=in_channels,out_channels=self.block_sizes[0]) self.resnet_block2 = ResNetBlock(in_channels=self.block_sizes[0],out_channels=self.block_sizes[1]) self.resnet_block3 = ResNetBlock(in_channels=self.block_sizes[1],out_channels=self.block_sizes[2]) self.global_pool = nn.AdaptiveAvgPool1d(1) self.fc1 = nn.Linear(self.block_sizes[2], num_classes) Results that model is showing are rather confusing, because the model is stuck at 47% training accuracy, and around 46% testing accuracy. What I tried was adding nn.Dropout(0.2) after every resnet block but it didn't help at all. I tried all sorts of combinations for learning_rate, from 0.01 to 0.00001, and I tried various batch_sizes ([32,64,128,256]), but I am stuck. Any help would be well appreciated. Continue reading...
