Decoding sensor space data with generalization across time and conditions#

This example runs the analysis described in [1]. It illustrates how one can fit a linear classifier to identify a discriminatory topography at a given time instant and subsequently assess whether this linear model can accurately predict all of the time samples of a second set of conditions.

# Authors: Jean-Remi King <jeanremi.king@gmail.com>
#          Alexandre Gramfort <alexandre.gramfort@inria.fr>
#          Denis Engemann <denis.engemann@gmail.com>
#
# License: BSD-3-Clause
# Copyright the MNE-Python contributors.
import matplotlib.pyplot as plt
from sklearn.linear_model import LogisticRegression
from sklearn.pipeline import make_pipeline
from sklearn.preprocessing import StandardScaler

import mne
from mne.datasets import sample
from mne.decoding import GeneralizingEstimator

print(__doc__)

# Preprocess data
data_path = sample.data_path()
# Load and filter data, set up epochs
meg_path = data_path / "MEG" / "sample"
raw_fname = meg_path / "sample_audvis_filt-0-40_raw.fif"
events_fname = meg_path / "sample_audvis_filt-0-40_raw-eve.fif"
raw = mne.io.read_raw_fif(raw_fname, preload=True)
picks = mne.pick_types(raw.info, meg=True, exclude="bads")  # Pick MEG channels
raw.filter(1.0, 30.0, fir_design="firwin")  # Band pass filtering signals
events = mne.read_events(events_fname)
event_id = {
    "Auditory/Left": 1,
    "Auditory/Right": 2,
    "Visual/Left": 3,
    "Visual/Right": 4,
}
tmin = -0.050
tmax = 0.400
# decimate to make the example faster to run, but then use verbose='error' in
# the Epochs constructor to suppress warning about decimation causing aliasing
decim = 2
epochs = mne.Epochs(
    raw,
    events,
    event_id=event_id,
    tmin=tmin,
    tmax=tmax,
    proj=True,
    picks=picks,
    baseline=None,
    preload=True,
    reject=dict(mag=5e-12),
    decim=decim,
    verbose="error",
)
Opening raw data file /home/circleci/mne_data/MNE-sample-data/MEG/sample/sample_audvis_filt-0-40_raw.fif...
    Read a total of 4 projection items:
        PCA-v1 (1 x 102)  idle
        PCA-v2 (1 x 102)  idle
        PCA-v3 (1 x 102)  idle
        Average EEG reference (1 x 60)  idle
    Range : 6450 ... 48149 =     42.956 ...   320.665 secs
Ready.
Reading 0 ... 41699  =      0.000 ...   277.709 secs...
Filtering raw data in 1 contiguous segment
Setting up band-pass filter from 1 - 30 Hz

FIR filter parameters
---------------------
Designing a one-pass, zero-phase, non-causal bandpass filter:
- Windowed time-domain design (firwin) method
- Hamming window with 0.0194 passband ripple and 53 dB stopband attenuation
- Lower passband edge: 1.00
- Lower transition bandwidth: 1.00 Hz (-6 dB cutoff frequency: 0.50 Hz)
- Upper passband edge: 30.00 Hz
- Upper transition bandwidth: 7.50 Hz (-6 dB cutoff frequency: 33.75 Hz)
- Filter length: 497 samples (3.310 s)

[Parallel(n_jobs=1)]: Done  17 tasks      | elapsed:    0.0s
[Parallel(n_jobs=1)]: Done  71 tasks      | elapsed:    0.1s
[Parallel(n_jobs=1)]: Done 161 tasks      | elapsed:    0.3s
[Parallel(n_jobs=1)]: Done 287 tasks      | elapsed:    0.6s

We will train the classifier on all left visual vs auditory trials and test on all right visual vs auditory trials.

clf = make_pipeline(
    StandardScaler(),
    LogisticRegression(solver="liblinear"),  # liblinear is faster than lbfgs
)
time_gen = GeneralizingEstimator(clf, scoring="roc_auc", n_jobs=None, verbose=True)

# Fit classifiers on the epochs where the stimulus was presented to the left.
# Note that the experimental condition y indicates auditory or visual
time_gen.fit(X=epochs["Left"].get_data(copy=False), y=epochs["Left"].events[:, 2] > 2)
  0%|          | Fitting GeneralizingEstimator : 0/35 [00:00<?,       ?it/s]
  6%|▌         | Fitting GeneralizingEstimator : 2/35 [00:00<00:00,   55.05it/s]
 11%|█▏        | Fitting GeneralizingEstimator : 4/35 [00:00<00:00,   57.18it/s]
 20%|██        | Fitting GeneralizingEstimator : 7/35 [00:00<00:00,   67.49it/s]
 26%|██▌       | Fitting GeneralizingEstimator : 9/35 [00:00<00:00,   65.37it/s]
 34%|███▍      | Fitting GeneralizingEstimator : 12/35 [00:00<00:00,   70.51it/s]
 43%|████▎     | Fitting GeneralizingEstimator : 15/35 [00:00<00:00,   73.98it/s]
 51%|█████▏    | Fitting GeneralizingEstimator : 18/35 [00:00<00:00,   76.45it/s]
 60%|██████    | Fitting GeneralizingEstimator : 21/35 [00:00<00:00,   78.31it/s]
 69%|██████▊   | Fitting GeneralizingEstimator : 24/35 [00:00<00:00,   79.77it/s]
 74%|███████▍  | Fitting GeneralizingEstimator : 26/35 [00:00<00:00,   76.65it/s]
 83%|████████▎ | Fitting GeneralizingEstimator : 29/35 [00:00<00:00,   78.08it/s]
 89%|████████▊ | Fitting GeneralizingEstimator : 31/35 [00:00<00:00,   76.08it/s]
 97%|█████████▋| Fitting GeneralizingEstimator : 34/35 [00:00<00:00,   77.40it/s]
100%|██████████| Fitting GeneralizingEstimator : 35/35 [00:00<00:00,   77.48it/s]

Score on the epochs where the stimulus was presented to the right.

scores = time_gen.score(
    X=epochs["Right"].get_data(copy=False), y=epochs["Right"].events[:, 2] > 2
)
  0%|          | Scoring GeneralizingEstimator : 0/1225 [00:00<?,       ?it/s]
  1%|          | Scoring GeneralizingEstimator : 12/1225 [00:00<00:03,  325.18it/s]
  2%|▏         | Scoring GeneralizingEstimator : 22/1225 [00:00<00:03,  311.27it/s]
  3%|▎         | Scoring GeneralizingEstimator : 33/1225 [00:00<00:03,  316.44it/s]
  4%|▎         | Scoring GeneralizingEstimator : 45/1225 [00:00<00:03,  326.78it/s]
  5%|▍         | Scoring GeneralizingEstimator : 57/1225 [00:00<00:03,  333.17it/s]
  6%|▌         | Scoring GeneralizingEstimator : 68/1225 [00:00<00:03,  330.47it/s]
  6%|▋         | Scoring GeneralizingEstimator : 79/1225 [00:00<00:03,  329.61it/s]
  7%|▋         | Scoring GeneralizingEstimator : 91/1225 [00:00<00:03,  331.76it/s]
  8%|▊         | Scoring GeneralizingEstimator : 103/1225 [00:00<00:03,  334.87it/s]
  9%|▉         | Scoring GeneralizingEstimator : 114/1225 [00:00<00:03,  333.67it/s]
 10%|█         | Scoring GeneralizingEstimator : 126/1225 [00:00<00:03,  336.20it/s]
 11%|█         | Scoring GeneralizingEstimator : 137/1225 [00:00<00:03,  335.03it/s]
 12%|█▏        | Scoring GeneralizingEstimator : 149/1225 [00:00<00:03,  336.09it/s]
 13%|█▎        | Scoring GeneralizingEstimator : 160/1225 [00:00<00:03,  335.12it/s]
 14%|█▍        | Scoring GeneralizingEstimator : 172/1225 [00:00<00:03,  336.92it/s]
 15%|█▌        | Scoring GeneralizingEstimator : 185/1225 [00:00<00:03,  338.84it/s]
 16%|█▌        | Scoring GeneralizingEstimator : 197/1225 [00:00<00:03,  340.28it/s]
 17%|█▋        | Scoring GeneralizingEstimator : 209/1225 [00:00<00:02,  341.48it/s]
 18%|█▊        | Scoring GeneralizingEstimator : 221/1225 [00:00<00:02,  342.54it/s]
 19%|█▉        | Scoring GeneralizingEstimator : 233/1225 [00:00<00:02,  343.54it/s]
 20%|██        | Scoring GeneralizingEstimator : 245/1225 [00:00<00:02,  344.49it/s]
 21%|██        | Scoring GeneralizingEstimator : 257/1225 [00:00<00:02,  345.33it/s]
 22%|██▏       | Scoring GeneralizingEstimator : 269/1225 [00:00<00:02,  346.08it/s]
 23%|██▎       | Scoring GeneralizingEstimator : 280/1225 [00:00<00:02,  344.65it/s]
 24%|██▍       | Scoring GeneralizingEstimator : 293/1225 [00:00<00:02,  346.56it/s]
 25%|██▍       | Scoring GeneralizingEstimator : 305/1225 [00:00<00:02,  347.14it/s]
 26%|██▌       | Scoring GeneralizingEstimator : 317/1225 [00:00<00:02,  347.69it/s]
 27%|██▋       | Scoring GeneralizingEstimator : 328/1225 [00:00<00:02,  346.21it/s]
 28%|██▊       | Scoring GeneralizingEstimator : 340/1225 [00:00<00:02,  346.74it/s]
 29%|██▉       | Scoring GeneralizingEstimator : 353/1225 [00:01<00:02,  347.22it/s]
 30%|██▉       | Scoring GeneralizingEstimator : 365/1225 [00:01<00:02,  347.76it/s]
 31%|███       | Scoring GeneralizingEstimator : 376/1225 [00:01<00:02,  346.46it/s]
 32%|███▏      | Scoring GeneralizingEstimator : 388/1225 [00:01<00:02,  346.99it/s]
 33%|███▎      | Scoring GeneralizingEstimator : 400/1225 [00:01<00:02,  347.53it/s]
 34%|███▎      | Scoring GeneralizingEstimator : 412/1225 [00:01<00:02,  348.04it/s]
 35%|███▍      | Scoring GeneralizingEstimator : 423/1225 [00:01<00:02,  346.76it/s]
 36%|███▌      | Scoring GeneralizingEstimator : 435/1225 [00:01<00:02,  347.26it/s]
 36%|███▋      | Scoring GeneralizingEstimator : 447/1225 [00:01<00:02,  347.75it/s]
 37%|███▋      | Scoring GeneralizingEstimator : 459/1225 [00:01<00:02,  346.74it/s]
 38%|███▊      | Scoring GeneralizingEstimator : 471/1225 [00:01<00:02,  347.21it/s]
 39%|███▉      | Scoring GeneralizingEstimator : 483/1225 [00:01<00:02,  347.72it/s]
 40%|████      | Scoring GeneralizingEstimator : 494/1225 [00:01<00:02,  346.53it/s]
 41%|████▏     | Scoring GeneralizingEstimator : 506/1225 [00:01<00:02,  347.03it/s]
 42%|████▏     | Scoring GeneralizingEstimator : 518/1225 [00:01<00:02,  347.50it/s]
 43%|████▎     | Scoring GeneralizingEstimator : 530/1225 [00:01<00:01,  347.99it/s]
 44%|████▍     | Scoring GeneralizingEstimator : 541/1225 [00:01<00:01,  346.70it/s]
 45%|████▌     | Scoring GeneralizingEstimator : 553/1225 [00:01<00:01,  346.62it/s]
 46%|████▌     | Scoring GeneralizingEstimator : 565/1225 [00:01<00:01,  347.11it/s]
 47%|████▋     | Scoring GeneralizingEstimator : 577/1225 [00:01<00:01,  347.55it/s]
 48%|████▊     | Scoring GeneralizingEstimator : 589/1225 [00:01<00:01,  347.96it/s]
 49%|████▉     | Scoring GeneralizingEstimator : 601/1225 [00:01<00:01,  347.31it/s]
 50%|█████     | Scoring GeneralizingEstimator : 613/1225 [00:01<00:01,  347.79it/s]
 51%|█████     | Scoring GeneralizingEstimator : 625/1225 [00:01<00:01,  348.16it/s]
 52%|█████▏    | Scoring GeneralizingEstimator : 636/1225 [00:01<00:01,  347.01it/s]
 53%|█████▎    | Scoring GeneralizingEstimator : 648/1225 [00:01<00:01,  347.40it/s]
 54%|█████▍    | Scoring GeneralizingEstimator : 660/1225 [00:01<00:01,  347.86it/s]
 55%|█████▍    | Scoring GeneralizingEstimator : 672/1225 [00:01<00:01,  347.47it/s]
 56%|█████▌    | Scoring GeneralizingEstimator : 684/1225 [00:01<00:01,  347.93it/s]
 57%|█████▋    | Scoring GeneralizingEstimator : 696/1225 [00:02<00:01,  348.30it/s]
 57%|█████▋    | Scoring GeneralizingEstimator : 703/1225 [00:02<00:01,  340.94it/s]
 58%|█████▊    | Scoring GeneralizingEstimator : 710/1225 [00:02<00:01,  331.00it/s]
 59%|█████▉    | Scoring GeneralizingEstimator : 720/1225 [00:02<00:01,  329.11it/s]
 60%|█████▉    | Scoring GeneralizingEstimator : 732/1225 [00:02<00:01,  330.52it/s]
 61%|██████    | Scoring GeneralizingEstimator : 744/1225 [00:02<00:01,  331.82it/s]
 62%|██████▏   | Scoring GeneralizingEstimator : 755/1225 [00:02<00:01,  330.80it/s]
 62%|██████▏   | Scoring GeneralizingEstimator : 764/1225 [00:02<00:01,  327.48it/s]
 63%|██████▎   | Scoring GeneralizingEstimator : 776/1225 [00:02<00:01,  328.98it/s]
 64%|██████▍   | Scoring GeneralizingEstimator : 788/1225 [00:02<00:01,  330.29it/s]
 65%|██████▌   | Scoring GeneralizingEstimator : 799/1225 [00:02<00:01,  329.99it/s]
 66%|██████▌   | Scoring GeneralizingEstimator : 811/1225 [00:02<00:01,  330.52it/s]
 67%|██████▋   | Scoring GeneralizingEstimator : 823/1225 [00:02<00:01,  331.83it/s]
 68%|██████▊   | Scoring GeneralizingEstimator : 834/1225 [00:02<00:01,  331.50it/s]
 69%|██████▉   | Scoring GeneralizingEstimator : 845/1225 [00:02<00:01,  331.25it/s]
 70%|██████▉   | Scoring GeneralizingEstimator : 857/1225 [00:02<00:01,  332.50it/s]
 71%|███████   | Scoring GeneralizingEstimator : 869/1225 [00:02<00:01,  333.57it/s]
 72%|███████▏  | Scoring GeneralizingEstimator : 881/1225 [00:02<00:01,  333.96it/s]
 73%|███████▎  | Scoring GeneralizingEstimator : 894/1225 [00:02<00:00,  336.41it/s]
 74%|███████▍  | Scoring GeneralizingEstimator : 906/1225 [00:02<00:00,  337.42it/s]
 75%|███████▍  | Scoring GeneralizingEstimator : 918/1225 [00:02<00:00,  338.37it/s]
 76%|███████▌  | Scoring GeneralizingEstimator : 930/1225 [00:02<00:00,  339.19it/s]
 77%|███████▋  | Scoring GeneralizingEstimator : 943/1225 [00:02<00:00,  340.93it/s]
 78%|███████▊  | Scoring GeneralizingEstimator : 956/1225 [00:02<00:00,  343.19it/s]
 79%|███████▉  | Scoring GeneralizingEstimator : 970/1225 [00:02<00:00,  346.79it/s]
 80%|████████  | Scoring GeneralizingEstimator : 985/1225 [00:02<00:00,  351.75it/s]
 82%|████████▏ | Scoring GeneralizingEstimator : 1001/1225 [00:02<00:00,  356.25it/s]
 83%|████████▎ | Scoring GeneralizingEstimator : 1015/1225 [00:02<00:00,  359.22it/s]
 84%|████████▍ | Scoring GeneralizingEstimator : 1030/1225 [00:02<00:00,  363.51it/s]
 85%|████████▌ | Scoring GeneralizingEstimator : 1045/1225 [00:03<00:00,  367.61it/s]
 86%|████████▋ | Scoring GeneralizingEstimator : 1059/1225 [00:03<00:00,  369.96it/s]
 88%|████████▊ | Scoring GeneralizingEstimator : 1074/1225 [00:03<00:00,  373.72it/s]
 89%|████████▉ | Scoring GeneralizingEstimator : 1088/1225 [00:03<00:00,  375.75it/s]
 90%|████████▉ | Scoring GeneralizingEstimator : 1102/1225 [00:03<00:00,  377.76it/s]
 91%|█████████ | Scoring GeneralizingEstimator : 1115/1225 [00:03<00:00,  378.10it/s]
 92%|█████████▏| Scoring GeneralizingEstimator : 1128/1225 [00:03<00:00,  378.45it/s]
 93%|█████████▎| Scoring GeneralizingEstimator : 1142/1225 [00:03<00:00,  380.29it/s]
 94%|█████████▍| Scoring GeneralizingEstimator : 1156/1225 [00:03<00:00,  382.02it/s]
 95%|█████████▌| Scoring GeneralizingEstimator : 1169/1225 [00:03<00:00,  380.85it/s]
 96%|█████████▋| Scoring GeneralizingEstimator : 1181/1225 [00:03<00:00,  379.57it/s]
 97%|█████████▋| Scoring GeneralizingEstimator : 1194/1225 [00:03<00:00,  379.87it/s]
 99%|█████████▊| Scoring GeneralizingEstimator : 1208/1225 [00:03<00:00,  381.60it/s]
100%|█████████▉| Scoring GeneralizingEstimator : 1221/1225 [00:03<00:00,  381.81it/s]
100%|██████████| Scoring GeneralizingEstimator : 1225/1225 [00:03<00:00,  354.54it/s]

Plot

fig, ax = plt.subplots(layout="constrained")
im = ax.matshow(
    scores,
    vmin=0,
    vmax=1.0,
    cmap="RdBu_r",
    origin="lower",
    extent=epochs.times[[0, -1, 0, -1]],
)
ax.axhline(0.0, color="k")
ax.axvline(0.0, color="k")
ax.xaxis.set_ticks_position("bottom")
ax.set_xlabel(
    'Condition: "Right"\nTesting Time (s)',
)
ax.set_ylabel('Condition: "Left"\nTraining Time (s)')
ax.set_title("Generalization across time and condition", fontweight="bold")
fig.colorbar(im, ax=ax, label="Performance (ROC AUC)")
plt.show()
Generalization across time and condition

References#

Total running time of the script: (0 minutes 5.839 seconds)

Gallery generated by Sphinx-Gallery