Skip to content

Instantly share code, notes, and snippets.

@iso-p

iso-p/CHB_Scalp_Seizure.ipynb

Forked from robintibor/CHB_Scalp_Seizure.ipynb
Created September 4, 2020 08:13
Show Gist options
  • Save iso-p/d34b62ab8e397970883b07163330fef2 to your computer and use it in GitHub Desktop.
Save iso-p/d34b62ab8e397970883b07163330fef2 to your computer and use it in GitHub Desktop.
Download ZIP
Braindecode on CHB-MIT Scalp EEG Database
Display the source blob
Display the rendered blob
Raw
CHB_Scalp_Seizure.ipynb
{
"cells": [
{
"cell_type": "code",
"execution_count": 1,
"metadata": {
"nbsphinx": "hidden"
},
"outputs": [],
"source": [
"%load_ext autoreload\n",
"%autoreload 2"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# CHB MIT Scalp Seizure Dataset"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"1. First download dataset from https://www.physionet.org/pn6/chbmit/ (for this notebook only chb01 is needed)\n",
"2. Change the `base_path` in the code below"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Load Data"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"import glob\n",
"import os.path\n",
"subject_id = 1\n",
"base_path = \"/data/schirrmr/physionet.org/physiobank/database/chbmit/\"\n",
"edf_file_names = sorted(glob.glob(os.path.join(base_path, \"chb{:02d}/*.edf\".format(subject_id))))\n",
"summary_file = os.path.join(base_path, \"chb{:02d}/chb{:02d}-summary.txt\".format(subject_id, subject_id))\n",
"\n",
"\n",
"summary_content = open(summary_file,'r').read()"
]
},
{
"cell_type": "code",
"execution_count": 3,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"import re\n",
"import mne\n",
"import numpy as np\n",
"def extract_data_and_labels(edf_filename, summary_text):\n",
" folder, basename = os.path.split(edf_filename)\n",
" \n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n",
" X = edf.get_data().astype(np.float32) * 1e6 # to mV\n",
" y = np.zeros(X.shape[1], dtype=np.int64)\n",
" i_text_start = summary_text.index(basename)\n",
"\n",
" if 'File Name' in summary_text[i_text_start:]:\n",
" i_text_stop = summary_text.index('File Name', i_text_start)\n",
" else:\n",
" i_text_stop = len(summary_text)\n",
" assert i_text_stop > i_text_start\n",
"\n",
" file_text = summary_text[i_text_start:i_text_stop]\n",
" if 'Seizure Start' in file_text:\n",
" start_sec = int(re.search(r\"Seizure Start Time: ([0-9]*) seconds\", file_text).group(1))\n",
" end_sec = int(re.search(r\"Seizure End Time: ([0-9]*) seconds\", file_text).group(1))\n",
" i_seizure_start = int(round(start_sec * edf.info['sfreq']))\n",
" i_seizure_stop = int(round((end_sec + 1) * edf.info['sfreq']))\n",
" y[i_seizure_start:i_seizure_stop] = 1\n",
" assert X.shape[1] == len(y)\n",
" return X,y"
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_01.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_02.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_03.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_04.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_05.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_06.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_07.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_08.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_09.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_10.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_11.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_12.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_13.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_14.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_15.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_16.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_17.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_18.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_19.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_20.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_21.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_22.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_23.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_24.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_25.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_26.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_27.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n",
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_29.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n",
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_30.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_31.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_32.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_33.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_34.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_36.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_37.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_38.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_39.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_40.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_41.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_42.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_43.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Extracting edf Parameters from /data/schirrmr/physionet.org/physiobank/database/chbmit/chb01/chb01_46.edf...\n",
"EDF file detected\n",
"Setting channel info structure...\n",
"Created Raw.info structure...\n",
"Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
"Ready.\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"<ipython-input-3-ae37fa0d04aa>:7: RuntimeWarning: Channel names are not unique, found duplicates for: {'T8-P8'}. Applying running numbers for duplicates.\n",
" edf = mne.io.read_raw_edf(edf_filename,stim_channel=None)\n"
]
}
],
"source": [
"all_X = []\n",
"all_y = []\n",
"for edf_file_name in edf_file_names:\n",
" X, y = extract_data_and_labels(edf_file_name, summary_content)\n",
" all_X.append(X)\n",
" all_y.append(y)"
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {},
"outputs": [],
"source": [
"dummy_X = all_X#[x[:,:2000] for x in all_X]\n",
"\n",
"dummy_y = all_y#[y[:2000] for y in all_y]\n",
"\n",
"from braindecode.datautil.signal_target import SignalAndTarget\n",
"from braindecode.datautil.splitters import split_into_two_sets\n",
"whole_set = SignalAndTarget(dummy_X, dummy_y)\n",
"train_set, test_set = split_into_two_sets(whole_set,0.5)\n",
"train_set, valid_set = split_into_two_sets(train_set, 0.7)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Create model"
]
},
{
"cell_type": "code",
"execution_count": 6,
"metadata": {},
"outputs": [],
"source": [
"from braindecode.models.shallow_fbcsp import ShallowFBCSPNet\n",
"from braindecode.models.deep4 import Deep4Net\n",
"from torch import nn\n",
"from braindecode.torch_ext.util import set_random_seeds\n",
"from braindecode.models.util import to_dense_prediction_model\n",
"\n",
"# Set if you want to use GPU\n",
"# You can also use torch.cuda.is_available() to determine if cuda is available on your machine.\n",
"cuda = True\n",
"set_random_seeds(seed=20170629, cuda=cuda)\n",
"\n",
"# This will determine how many crops are processed in parallel\n",
"input_time_length = 1200\n",
"n_classes = 2\n",
"in_chans = train_set.X[0].shape[0]\n",
"# final_conv_length determines the size of the receptive field of the ConvNet\n",
"#model = ShallowFBCSPNet(in_chans=in_chans, n_classes=n_classes, input_time_length=input_time_length,\n",
"# final_conv_length=12).create_network()\n",
"model = Deep4Net(in_chans=in_chans, n_classes=n_classes, input_time_length=input_time_length,\n",
" final_conv_length=2, stride_before_pool=True).create_network()\n",
"to_dense_prediction_model(model)\n",
"\n",
"if cuda:\n",
" model.cuda()"
]
},
{
"cell_type": "code",
"execution_count": 7,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"from torch import optim\n",
"\n",
"optimizer = optim.Adam(model.parameters())"
]
},
{
"cell_type": "code",
"execution_count": 8,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"519 predictions per input/trial\n"
]
}
],
"source": [
"from braindecode.torch_ext.util import np_to_var\n",
"# determine output size\n",
"test_input = np_to_var(np.ones((2, in_chans, input_time_length, 1), dtype=np.float32))\n",
"if cuda:\n",
" test_input = test_input.cuda()\n",
"out = model(test_input)\n",
"n_preds_per_input = out.cpu().data.numpy().shape[2]\n",
"print(\"{:d} predictions per input/trial\".format(n_preds_per_input))"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Monitor that computes multiple metrics\n",
"\n",
"You can adapt it to compute other metrics you are interested in.."
]
},
{
"cell_type": "code",
"execution_count": 9,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"from braindecode.experiments.monitors import compute_preds_per_trial_for_set\n",
"\n",
"from sklearn.metrics import roc_auc_score\n",
"\n",
"\n",
"class SeizureMonitor(object):\n",
" \"\"\"\n",
" Compute trialwise misclasses from predictions for crops.\n",
" \n",
" Parameters\n",
" ----------\n",
" input_time_length: int\n",
" Temporal length of one input to the model.\n",
" \"\"\"\n",
" def __init__(self, input_time_length=None):\n",
" self.input_time_length = input_time_length\n",
"\n",
" def monitor_epoch(self,):\n",
" return\n",
"\n",
" def monitor_set(self, setname, all_preds, all_losses,\n",
" all_batch_sizes, all_targets, dataset):\n",
" \"\"\"Assuming one hot encoding for now\"\"\"\n",
" assert self.input_time_length is not None, \"Need to know input time length...\"\n",
" # this will be timeseries of predictions\n",
" # for each trial\n",
" preds_per_trial = compute_preds_per_trial_for_set(all_preds, self.input_time_length, \n",
" dataset)\n",
" seizure_preds = []\n",
" all_preds = []\n",
" all_y = []\n",
" for i_trial in range(len(preds_per_trial)):\n",
" this_y = dataset.y[i_trial]\n",
" this_preds = preds_per_trial[i_trial]\n",
" this_preds = np.exp(this_preds[1])\n",
" n_missing_preds = len(this_y) - len(this_preds)\n",
" this_preds = np.concatenate((np.zeros(n_missing_preds, dtype=this_preds.dtype),\n",
" this_preds))\n",
" all_preds.extend(this_preds)\n",
" all_y.extend(this_y)\n",
" if np.any(this_y == 1):\n",
" seizure_preds.append(this_preds[this_y == 1])\n",
" if len(seizure_preds) > 0:\n",
" max_seiz_preds = np.array([np.max(p) for p in seizure_preds])\n",
" sensitivity = np.mean(max_seiz_preds > 0.5)\n",
" else:\n",
" sensitivity = np.nan\n",
" \n",
" sensitivity_name = \"{:s}_sensitivity\".format(setname)\n",
" if len(np.unique(all_y)) > 1:\n",
" auc = roc_auc_score(all_y, all_preds)\n",
" else:\n",
" auc = np.nan\n",
" auc_name = \"{:s}_auc\".format(setname)\n",
" return {sensitivity_name: float(sensitivity),\n",
" auc_name: float(auc)}"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Setup Experiment and Run"
]
},
{
"cell_type": "code",
"execution_count": 10,
"metadata": {
"collapsed": true
},
"outputs": [],
"source": [
"from braindecode.torch_ext.losses import log_categorical_crossentropy\n",
"from braindecode.experiments.experiment import Experiment\n",
"from braindecode.datautil.iterators import CropsFromTrialsIterator\n",
"from braindecode.experiments.monitors import RuntimeMonitor, LossMonitor, CroppedTrialMisclassMonitor, MisclassMonitor\n",
"from braindecode.experiments.stopcriteria import MaxEpochs\n",
"import torch.nn.functional as F\n",
"import torch as th\n",
"from braindecode.torch_ext.modules import Expression\n",
"# Iterator is used to iterate over datasets both for training\n",
"# and evaluation\n",
"iterator = CropsFromTrialsIterator(batch_size=32,input_time_length=input_time_length,\n",
" n_preds_per_input=n_preds_per_input)\n",
"\n",
"# Loss function takes predictions as they come out of the network and the targets\n",
"# and returns a loss\n",
"loss_function = lambda preds, targets: log_categorical_crossentropy(preds, targets)\n",
"# Could be used to apply some constraint on the models, then should be object\n",
"# with apply method that accepts a module\n",
"model_constraint = None\n",
"# Monitors log the training progress\n",
"monitors = [LossMonitor(), MisclassMonitor(col_suffix='misclass'),\n",
" SeizureMonitor(input_time_length),\n",
" RuntimeMonitor(),]\n",
"# Stop criterion determines when the first stop happens\n",
"stop_criterion = MaxEpochs(5)\n",
"exp = Experiment(model, train_set, valid_set, test_set, iterator, loss_function, optimizer, model_constraint,\n",
" monitors, stop_criterion, remember_best_column='valid_misclass',\n",
" run_after_early_stop=True, batch_modifier=None, cuda=cuda)"
]
},
{
"cell_type": "code",
"execution_count": 11,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"2017-10-03 17:25:58,068 INFO : Run until first stop...\n",
"2017-10-03 17:27:30,965 INFO : Epoch 0\n",
"2017-10-03 17:27:30,966 INFO : train_loss 0.74135\n",
"2017-10-03 17:27:30,967 INFO : valid_loss 1.46873\n",
"2017-10-03 17:27:30,968 INFO : test_loss 0.57399\n",
"2017-10-03 17:27:30,969 INFO : train_misclass 0.03220\n",
"2017-10-03 17:27:30,970 INFO : valid_misclass 0.01954\n",
"2017-10-03 17:27:30,971 INFO : test_misclass 0.02823\n",
"2017-10-03 17:27:30,972 INFO : train_sensitivity 1.00000\n",
"2017-10-03 17:27:30,973 INFO : train_auc 0.14261\n",
"2017-10-03 17:27:30,974 INFO : valid_sensitivity 1.00000\n",
"2017-10-03 17:27:30,975 INFO : valid_auc 0.11537\n",
"2017-10-03 17:27:30,976 INFO : test_sensitivity 1.00000\n",
"2017-10-03 17:27:30,977 INFO : test_auc 0.14126\n",
"2017-10-03 17:27:30,978 INFO : runtime 0.00000\n",
"2017-10-03 17:27:30,979 INFO : \n",
"2017-10-03 17:27:30,986 INFO : New best valid_misclass: 0.019545\n",
"2017-10-03 17:27:30,987 INFO : \n",
"2017-10-03 17:28:51,310 INFO : Time only for training updates: 75.69s\n",
"2017-10-03 17:30:23,441 INFO : Epoch 1\n",
"2017-10-03 17:30:23,443 INFO : train_loss 0.00781\n",
"2017-10-03 17:30:23,444 INFO : valid_loss 0.07895\n",
"2017-10-03 17:30:23,445 INFO : test_loss 0.00743\n",
"2017-10-03 17:30:23,445 INFO : train_misclass 0.00143\n",
"2017-10-03 17:30:23,446 INFO : valid_misclass 0.01001\n",
"2017-10-03 17:30:23,447 INFO : test_misclass 0.00126\n",
"2017-10-03 17:30:23,448 INFO : train_sensitivity 1.00000\n",
"2017-10-03 17:30:23,449 INFO : train_auc 0.92865\n",
"2017-10-03 17:30:23,449 INFO : valid_sensitivity 1.00000\n",
"2017-10-03 17:30:23,450 INFO : valid_auc 0.83478\n",
"2017-10-03 17:30:23,451 INFO : test_sensitivity 1.00000\n",
"2017-10-03 17:30:23,452 INFO : test_auc 0.92489\n",
"2017-10-03 17:30:23,453 INFO : runtime 173.24244\n",
"2017-10-03 17:30:23,453 INFO : \n",
"2017-10-03 17:30:23,458 INFO : New best valid_misclass: 0.010006\n",
"2017-10-03 17:30:23,459 INFO : \n",
"2017-10-03 17:31:44,107 INFO : Time only for training updates: 76.02s\n",
"2017-10-03 17:33:16,295 INFO : Epoch 2\n",
"2017-10-03 17:33:16,297 INFO : train_loss 0.00128\n",
"2017-10-03 17:33:16,298 INFO : valid_loss 0.04533\n",
"2017-10-03 17:33:16,299 INFO : test_loss 0.00424\n",
"2017-10-03 17:33:16,300 INFO : train_misclass 0.00021\n",
"2017-10-03 17:33:16,300 INFO : valid_misclass 0.00579\n",
"2017-10-03 17:33:16,301 INFO : test_misclass 0.00115\n",
"2017-10-03 17:33:16,302 INFO : train_sensitivity 1.00000\n",
"2017-10-03 17:33:16,303 INFO : train_auc 0.99708\n",
"2017-10-03 17:33:16,304 INFO : valid_sensitivity 1.00000\n",
"2017-10-03 17:33:16,304 INFO : valid_auc 0.95073\n",
"2017-10-03 17:33:16,305 INFO : test_sensitivity 1.00000\n",
"2017-10-03 17:33:16,306 INFO : test_auc 0.98772\n",
"2017-10-03 17:33:16,307 INFO : runtime 172.79666\n",
"2017-10-03 17:33:16,308 INFO : \n",
"2017-10-03 17:33:16,312 INFO : New best valid_misclass: 0.005794\n",
"2017-10-03 17:33:16,313 INFO : \n",
"2017-10-03 17:34:36,847 INFO : Time only for training updates: 75.93s\n",
"2017-10-03 17:36:09,110 INFO : Epoch 3\n",
"2017-10-03 17:36:09,112 INFO : train_loss 0.00071\n",
"2017-10-03 17:36:09,113 INFO : valid_loss 0.03611\n",
"2017-10-03 17:36:09,114 INFO : test_loss 0.00392\n",
"2017-10-03 17:36:09,114 INFO : train_misclass 0.00021\n",
"2017-10-03 17:36:09,115 INFO : valid_misclass 0.00533\n",
"2017-10-03 17:36:09,116 INFO : test_misclass 0.00100\n",
"2017-10-03 17:36:09,117 INFO : train_sensitivity 1.00000\n",
"2017-10-03 17:36:09,118 INFO : train_auc 0.99982\n",
"2017-10-03 17:36:09,119 INFO : valid_sensitivity 1.00000\n",
"2017-10-03 17:36:09,119 INFO : valid_auc 0.95862\n",
"2017-10-03 17:36:09,120 INFO : test_sensitivity 1.00000\n",
"2017-10-03 17:36:09,121 INFO : test_auc 0.99435\n",
"2017-10-03 17:36:09,122 INFO : runtime 172.74023\n",
"2017-10-03 17:36:09,123 INFO : \n",
"2017-10-03 17:36:09,126 INFO : New best valid_misclass: 0.005333\n",
"2017-10-03 17:36:09,127 INFO : \n",
"2017-10-03 17:37:29,685 INFO : Time only for training updates: 75.95s\n",
"2017-10-03 17:39:01,783 INFO : Epoch 4\n",
"2017-10-03 17:39:01,784 INFO : train_loss 0.00078\n",
"2017-10-03 17:39:01,785 INFO : valid_loss 0.05229\n",
"2017-10-03 17:39:01,786 INFO : test_loss 0.00317\n",
"2017-10-03 17:39:01,787 INFO : train_misclass 0.00019\n",
"2017-10-03 17:39:01,788 INFO : valid_misclass 0.00564\n",
"2017-10-03 17:39:01,789 INFO : test_misclass 0.00079\n",
"2017-10-03 17:39:01,789 INFO : train_sensitivity 1.00000\n",
"2017-10-03 17:39:01,790 INFO : train_auc 0.99982\n",
"2017-10-03 17:39:01,791 INFO : valid_sensitivity 1.00000\n",
"2017-10-03 17:39:01,792 INFO : valid_auc 0.95368\n",
"2017-10-03 17:39:01,793 INFO : test_sensitivity 1.00000\n",
"2017-10-03 17:39:01,793 INFO : test_auc 0.99381\n",
"2017-10-03 17:39:01,794 INFO : runtime 172.83778\n",
"2017-10-03 17:39:01,795 INFO : \n",
"2017-10-03 17:40:22,336 INFO : Time only for training updates: 75.95s\n",
"2017-10-03 17:41:54,442 INFO : Epoch 5\n",
"2017-10-03 17:41:54,443 INFO : train_loss 0.00117\n",
"2017-10-03 17:41:54,444 INFO : valid_loss 0.06295\n",
"2017-10-03 17:41:54,445 INFO : test_loss 0.00552\n",
"2017-10-03 17:41:54,446 INFO : train_misclass 0.00019\n",
"2017-10-03 17:41:54,447 INFO : valid_misclass 0.00617\n",
"2017-10-03 17:41:54,447 INFO : test_misclass 0.00123\n",
"2017-10-03 17:41:54,448 INFO : train_sensitivity 1.00000\n",
"2017-10-03 17:41:54,449 INFO : train_auc 0.99911\n",
"2017-10-03 17:41:54,450 INFO : valid_sensitivity 1.00000\n",
"2017-10-03 17:41:54,451 INFO : valid_auc 0.88697\n",
"2017-10-03 17:41:54,451 INFO : test_sensitivity 1.00000\n",
"2017-10-03 17:41:54,452 INFO : test_auc 0.98754\n",
"2017-10-03 17:41:54,453 INFO : runtime 172.65101\n",
"2017-10-03 17:41:54,454 INFO : \n",
"2017-10-03 17:41:54,455 INFO : Setup for second stop...\n",
"2017-10-03 17:41:54,459 INFO : Train loss to reach 0.00071\n",
"2017-10-03 17:41:54,459 INFO : Run until second stop...\n",
"2017-10-03 17:43:39,735 INFO : Epoch 4\n",
"2017-10-03 17:43:39,737 INFO : train_loss 0.01051\n",
"2017-10-03 17:43:39,738 INFO : valid_loss 0.03611\n",
"2017-10-03 17:43:39,739 INFO : test_loss 0.00392\n",
"2017-10-03 17:43:39,739 INFO : train_misclass 0.00163\n",
"2017-10-03 17:43:39,740 INFO : valid_misclass 0.00533\n",
"2017-10-03 17:43:39,741 INFO : test_misclass 0.00100\n",
"2017-10-03 17:43:39,742 INFO : train_sensitivity 1.00000\n",
"2017-10-03 17:43:39,743 INFO : train_auc 0.96786\n",
"2017-10-03 17:43:39,743 INFO : valid_sensitivity 1.00000\n",
"2017-10-03 17:43:39,744 INFO : valid_auc 0.95862\n",
"2017-10-03 17:43:39,745 INFO : test_sensitivity 1.00000\n",
"2017-10-03 17:43:39,746 INFO : test_auc 0.99435\n",
"2017-10-03 17:43:39,747 INFO : runtime 92.12272\n",
"2017-10-03 17:43:39,747 INFO : \n",
"2017-10-03 17:45:31,098 INFO : Time only for training updates: 104.97s\n",
"2017-10-03 17:47:16,342 INFO : Epoch 5\n",
"2017-10-03 17:47:16,344 INFO : train_loss 0.00552\n",
"2017-10-03 17:47:16,345 INFO : valid_loss 0.01261\n",
"2017-10-03 17:47:16,345 INFO : test_loss 0.00146\n",
"2017-10-03 17:47:16,346 INFO : train_misclass 0.00161\n",
"2017-10-03 17:47:16,347 INFO : valid_misclass 0.00357\n",
"2017-10-03 17:47:16,348 INFO : test_misclass 0.00034\n",
"2017-10-03 17:47:16,349 INFO : train_sensitivity 1.00000\n",
"2017-10-03 17:47:16,349 INFO : train_auc 0.99764\n",
"2017-10-03 17:47:16,350 INFO : valid_sensitivity 1.00000\n",
"2017-10-03 17:47:16,351 INFO : valid_auc 0.99753\n",
"2017-10-03 17:47:16,352 INFO : test_sensitivity 1.00000\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"2017-10-03 17:47:16,353 INFO : test_auc 0.98500\n",
"2017-10-03 17:47:16,353 INFO : runtime 216.63896\n",
"2017-10-03 17:47:16,354 INFO : \n",
"2017-10-03 17:49:07,741 INFO : Time only for training updates: 105.01s\n",
"2017-10-03 17:50:52,914 INFO : Epoch 6\n",
"2017-10-03 17:50:52,916 INFO : train_loss 0.00131\n",
"2017-10-03 17:50:52,917 INFO : valid_loss 0.00319\n",
"2017-10-03 17:50:52,918 INFO : test_loss 0.00525\n",
"2017-10-03 17:50:52,919 INFO : train_misclass 0.00031\n",
"2017-10-03 17:50:52,919 INFO : valid_misclass 0.00060\n",
"2017-10-03 17:50:52,920 INFO : test_misclass 0.00228\n",
"2017-10-03 17:50:52,921 INFO : train_sensitivity 1.00000\n",
"2017-10-03 17:50:52,922 INFO : train_auc 0.99926\n",
"2017-10-03 17:50:52,923 INFO : valid_sensitivity 1.00000\n",
"2017-10-03 17:50:52,923 INFO : valid_auc 0.99910\n",
"2017-10-03 17:50:52,924 INFO : test_sensitivity 1.00000\n",
"2017-10-03 17:50:52,925 INFO : test_auc 0.99784\n",
"2017-10-03 17:50:52,926 INFO : runtime 216.64343\n",
"2017-10-03 17:50:52,927 INFO : \n"
]
}
],
"source": [
"# need to setup python logging before to be able to see anything\n",
"import logging\n",
"import sys\n",
"logging.basicConfig(format='%(asctime)s %(levelname)s : %(message)s',\n",
" level=logging.DEBUG, stream=sys.stdout)\n",
"exp.run()"
]
}
],
"metadata": {
"celltoolbar": "Edit Metadata",
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.6.2"
}
},
"nbformat": 4,
"nbformat_minor": 2
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment