pbnsilva

net = Net()
net.to(device)

criterion = nn.CrossEntropyLoss()
optimizer = optim.Adam(net.parameters(), lr=0.001)

train_loss, val_loss = [], []
train_acc, val_acc = [], []
for epoch in range(500):
    train_running_loss = 0.0

pbnsilva

class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        filters_start = 32
        layer_filters = filters_start
        filters_growth = 32
        strides_start = 1
        strides_end = 2
        depth = 4
        n_blocks = 6

pbnsilva

class Conv2dBlock(nn.Module):
    def __init__(self, depth, layer_filters, filters_growth,
                 strides_start, strides_end, input_shape, first_layer=False):
        super(Conv2dBlock, self).__init__()
        layers = []
        c_in_channels = layer_filters
        for i in range(depth):
            if first_layer:
                layers.append(nn.Conv2d(in_channels=1,
                                        kernel_size=3,

pbnsilva

batch_size = 32

dataset = PhysioNetDataset(ref_file='training2017/REFERENCE.csv', data_dir='training2017')

train_size = int(0.7*len(dataset))
train_set, val_set = random_split(dataset, [train_size, len(dataset)-train_size])

train_loader = DataLoader(train_set, batch_size=batch_size, shuffle=True, num_workers=4)
val_loader = DataLoader(val_set, batch_size=batch_size, shuffle=True, num_workers=4)

pbnsilva

import os
import pandas as pd
from torch.utils.data import Dataset


class PhysioNetDataset(Dataset):
    def __init__(self, ref_file, data_dir):
        self.ref_file = ref_file
        self.data_dir = data_dir
        self.ref_frame = pd.read_csv(ref_file, names=['mat', 'label'])

pbnsilva

from scipy import signal

def spectrogram(data, fs=300, nperseg=64, noverlap=32):
    f, t, Sxx = signal.spectrogram(data, fs=fs, nperseg=nperseg, noverlap=noverlap)
    Sxx = np.transpose(Sxx, [0, 2, 1])
    Sxx = np.abs(Sxx)
    mask = Sxx > 0
    Sxx[mask] = np.log(Sxx[mask])
    return f, t, Sxx

pbnsilva

def zero_pad(data, length):
    extended = np.zeros(length)
    siglength = np.min([length, data.shape[0]])
    extended[:siglength] = data[:siglength]
    return extended

# plot a sample
data = sio.loadmat('training2017/A00001.mat')['val'][0]

# 61 seconds is the maximum length in our dataset

pbnsilva

import numpy as np
import scipy.io as sio
import matplotlib.pyplot as plt

# plot 10 seconds
secs = 10

# maximum length of ECG recording is 61 seconds
max_length = 61

pbnsilva

Keybase proof

I hereby claim:

• I am pbnsilva on github.
• I am pbnsilva (https://keybase.io/pbnsilva) on keybase.
• I have a public key whose fingerprint is 88AA DC61 1A86 0938 E6B0 4024 2196 9819 5B83 83A5

To claim this, I am signing this object:

pbnsilva

package main

import (
	"fmt"
	"math"
	"math/rand"

	"github.com/skelterjohn/go.matrix"
)