Note
Go to the end to download the full example code.
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: 0.5s
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]
3%|▎ | Fitting GeneralizingEstimator : 1/35 [00:00<00:01, 29.01it/s]
9%|▊ | Fitting GeneralizingEstimator : 3/35 [00:00<00:00, 44.27it/s]
14%|█▍ | Fitting GeneralizingEstimator : 5/35 [00:00<00:00, 49.37it/s]
20%|██ | Fitting GeneralizingEstimator : 7/35 [00:00<00:00, 51.79it/s]
26%|██▌ | Fitting GeneralizingEstimator : 9/35 [00:00<00:00, 53.41it/s]
31%|███▏ | Fitting GeneralizingEstimator : 11/35 [00:00<00:00, 54.48it/s]
40%|████ | Fitting GeneralizingEstimator : 14/35 [00:00<00:00, 60.11it/s]
49%|████▊ | Fitting GeneralizingEstimator : 17/35 [00:00<00:00, 64.30it/s]
54%|█████▍ | Fitting GeneralizingEstimator : 19/35 [00:00<00:00, 63.60it/s]
63%|██████▎ | Fitting GeneralizingEstimator : 22/35 [00:00<00:00, 66.54it/s]
69%|██████▊ | Fitting GeneralizingEstimator : 24/35 [00:00<00:00, 65.65it/s]
74%|███████▍ | Fitting GeneralizingEstimator : 26/35 [00:00<00:00, 64.93it/s]
80%|████████ | Fitting GeneralizingEstimator : 28/35 [00:00<00:00, 64.33it/s]
86%|████████▌ | Fitting GeneralizingEstimator : 30/35 [00:00<00:00, 63.82it/s]
94%|█████████▍| Fitting GeneralizingEstimator : 33/35 [00:00<00:00, 66.12it/s]
100%|██████████| Fitting GeneralizingEstimator : 35/35 [00:00<00:00, 66.27it/s]
100%|██████████| Fitting GeneralizingEstimator : 35/35 [00:00<00:00, 64.84it/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]
0%| | Scoring GeneralizingEstimator : 5/1225 [00:00<00:08, 145.03it/s]
1%| | Scoring GeneralizingEstimator : 13/1225 [00:00<00:06, 190.85it/s]
2%|▏ | Scoring GeneralizingEstimator : 20/1225 [00:00<00:06, 194.93it/s]
2%|▏ | Scoring GeneralizingEstimator : 27/1225 [00:00<00:06, 197.89it/s]
3%|▎ | Scoring GeneralizingEstimator : 35/1225 [00:00<00:05, 205.71it/s]
3%|▎ | Scoring GeneralizingEstimator : 42/1225 [00:00<00:05, 205.81it/s]
4%|▍ | Scoring GeneralizingEstimator : 49/1225 [00:00<00:05, 205.82it/s]
5%|▍ | Scoring GeneralizingEstimator : 56/1225 [00:00<00:05, 205.62it/s]
5%|▌ | Scoring GeneralizingEstimator : 64/1225 [00:00<00:05, 209.75it/s]
6%|▌ | Scoring GeneralizingEstimator : 71/1225 [00:00<00:05, 208.97it/s]
6%|▋ | Scoring GeneralizingEstimator : 79/1225 [00:00<00:05, 212.00it/s]
7%|▋ | Scoring GeneralizingEstimator : 87/1225 [00:00<00:05, 214.63it/s]
8%|▊ | Scoring GeneralizingEstimator : 95/1225 [00:00<00:05, 216.32it/s]
8%|▊ | Scoring GeneralizingEstimator : 102/1225 [00:00<00:05, 215.25it/s]
9%|▉ | Scoring GeneralizingEstimator : 110/1225 [00:00<00:05, 216.98it/s]
10%|▉ | Scoring GeneralizingEstimator : 117/1225 [00:00<00:05, 216.02it/s]
10%|█ | Scoring GeneralizingEstimator : 124/1225 [00:00<00:05, 215.22it/s]
11%|█ | Scoring GeneralizingEstimator : 132/1225 [00:00<00:05, 216.98it/s]
11%|█▏ | Scoring GeneralizingEstimator : 139/1225 [00:00<00:05, 216.13it/s]
12%|█▏ | Scoring GeneralizingEstimator : 150/1225 [00:00<00:04, 224.54it/s]
13%|█▎ | Scoring GeneralizingEstimator : 157/1225 [00:00<00:04, 223.09it/s]
13%|█▎ | Scoring GeneralizingEstimator : 165/1225 [00:00<00:04, 223.90it/s]
14%|█▍ | Scoring GeneralizingEstimator : 173/1225 [00:00<00:04, 224.43it/s]
15%|█▍ | Scoring GeneralizingEstimator : 180/1225 [00:00<00:04, 223.03it/s]
15%|█▌ | Scoring GeneralizingEstimator : 188/1225 [00:00<00:04, 223.74it/s]
16%|█▌ | Scoring GeneralizingEstimator : 195/1225 [00:00<00:04, 222.57it/s]
17%|█▋ | Scoring GeneralizingEstimator : 203/1225 [00:00<00:04, 223.43it/s]
17%|█▋ | Scoring GeneralizingEstimator : 211/1225 [00:00<00:04, 224.16it/s]
18%|█▊ | Scoring GeneralizingEstimator : 219/1225 [00:00<00:04, 224.78it/s]
18%|█▊ | Scoring GeneralizingEstimator : 226/1225 [00:01<00:04, 223.33it/s]
19%|█▉ | Scoring GeneralizingEstimator : 234/1225 [00:01<00:04, 224.17it/s]
20%|█▉ | Scoring GeneralizingEstimator : 241/1225 [00:01<00:04, 223.07it/s]
20%|██ | Scoring GeneralizingEstimator : 248/1225 [00:01<00:04, 221.79it/s]
21%|██ | Scoring GeneralizingEstimator : 255/1225 [00:01<00:04, 220.62it/s]
21%|██▏ | Scoring GeneralizingEstimator : 263/1225 [00:01<00:04, 221.56it/s]
22%|██▏ | Scoring GeneralizingEstimator : 270/1225 [00:01<00:04, 220.59it/s]
23%|██▎ | Scoring GeneralizingEstimator : 278/1225 [00:01<00:04, 221.39it/s]
23%|██▎ | Scoring GeneralizingEstimator : 285/1225 [00:01<00:04, 220.43it/s]
24%|██▍ | Scoring GeneralizingEstimator : 292/1225 [00:01<00:04, 219.52it/s]
24%|██▍ | Scoring GeneralizingEstimator : 300/1225 [00:01<00:04, 220.12it/s]
25%|██▌ | Scoring GeneralizingEstimator : 307/1225 [00:01<00:04, 219.16it/s]
26%|██▌ | Scoring GeneralizingEstimator : 315/1225 [00:01<00:04, 219.82it/s]
26%|██▋ | Scoring GeneralizingEstimator : 322/1225 [00:01<00:04, 218.96it/s]
27%|██▋ | Scoring GeneralizingEstimator : 329/1225 [00:01<00:04, 218.23it/s]
28%|██▊ | Scoring GeneralizingEstimator : 337/1225 [00:01<00:04, 219.20it/s]
28%|██▊ | Scoring GeneralizingEstimator : 344/1225 [00:01<00:04, 218.25it/s]
29%|██▊ | Scoring GeneralizingEstimator : 352/1225 [00:01<00:03, 219.20it/s]
29%|██▉ | Scoring GeneralizingEstimator : 359/1225 [00:01<00:03, 218.47it/s]
30%|██▉ | Scoring GeneralizingEstimator : 366/1225 [00:01<00:03, 217.83it/s]
31%|███ | Scoring GeneralizingEstimator : 374/1225 [00:01<00:03, 218.76it/s]
31%|███ | Scoring GeneralizingEstimator : 382/1225 [00:01<00:03, 219.59it/s]
32%|███▏ | Scoring GeneralizingEstimator : 392/1225 [00:01<00:03, 223.63it/s]
33%|███▎ | Scoring GeneralizingEstimator : 402/1225 [00:01<00:03, 227.44it/s]
34%|███▎ | Scoring GeneralizingEstimator : 412/1225 [00:01<00:03, 230.96it/s]
34%|███▍ | Scoring GeneralizingEstimator : 422/1225 [00:01<00:03, 234.33it/s]
35%|███▌ | Scoring GeneralizingEstimator : 434/1225 [00:01<00:03, 240.64it/s]
36%|███▋ | Scoring GeneralizingEstimator : 445/1225 [00:01<00:03, 244.95it/s]
37%|███▋ | Scoring GeneralizingEstimator : 457/1225 [00:01<00:03, 250.68it/s]
38%|███▊ | Scoring GeneralizingEstimator : 470/1225 [00:02<00:02, 257.51it/s]
39%|███▉ | Scoring GeneralizingEstimator : 481/1225 [00:02<00:02, 261.00it/s]
40%|████ | Scoring GeneralizingEstimator : 490/1225 [00:02<00:02, 261.23it/s]
41%|████ | Scoring GeneralizingEstimator : 500/1225 [00:02<00:02, 262.95it/s]
42%|████▏ | Scoring GeneralizingEstimator : 509/1225 [00:02<00:02, 263.14it/s]
42%|████▏ | Scoring GeneralizingEstimator : 518/1225 [00:02<00:02, 263.30it/s]
43%|████▎ | Scoring GeneralizingEstimator : 527/1225 [00:02<00:02, 263.39it/s]
44%|████▎ | Scoring GeneralizingEstimator : 535/1225 [00:02<00:02, 261.90it/s]
44%|████▍ | Scoring GeneralizingEstimator : 544/1225 [00:02<00:02, 262.07it/s]
45%|████▌ | Scoring GeneralizingEstimator : 553/1225 [00:02<00:02, 262.18it/s]
46%|████▌ | Scoring GeneralizingEstimator : 563/1225 [00:02<00:02, 263.60it/s]
47%|████▋ | Scoring GeneralizingEstimator : 572/1225 [00:02<00:02, 263.62it/s]
48%|████▊ | Scoring GeneralizingEstimator : 582/1225 [00:02<00:02, 265.24it/s]
48%|████▊ | Scoring GeneralizingEstimator : 594/1225 [00:02<00:02, 269.64it/s]
49%|████▉ | Scoring GeneralizingEstimator : 604/1225 [00:02<00:02, 270.93it/s]
50%|█████ | Scoring GeneralizingEstimator : 613/1225 [00:02<00:02, 270.63it/s]
51%|█████ | Scoring GeneralizingEstimator : 623/1225 [00:02<00:02, 271.89it/s]
52%|█████▏ | Scoring GeneralizingEstimator : 632/1225 [00:02<00:02, 271.62it/s]
52%|█████▏ | Scoring GeneralizingEstimator : 643/1225 [00:02<00:02, 274.15it/s]
53%|█████▎ | Scoring GeneralizingEstimator : 653/1225 [00:02<00:02, 275.22it/s]
54%|█████▍ | Scoring GeneralizingEstimator : 663/1225 [00:02<00:02, 276.12it/s]
55%|█████▌ | Scoring GeneralizingEstimator : 675/1225 [00:02<00:01, 279.95it/s]
56%|█████▌ | Scoring GeneralizingEstimator : 686/1225 [00:02<00:01, 282.21it/s]
57%|█████▋ | Scoring GeneralizingEstimator : 696/1225 [00:02<00:01, 282.88it/s]
58%|█████▊ | Scoring GeneralizingEstimator : 708/1225 [00:02<00:01, 286.46it/s]
59%|█████▉ | Scoring GeneralizingEstimator : 720/1225 [00:02<00:01, 289.72it/s]
60%|█████▉ | Scoring GeneralizingEstimator : 731/1225 [00:02<00:01, 291.48it/s]
61%|██████ | Scoring GeneralizingEstimator : 744/1225 [00:02<00:01, 296.12it/s]
62%|██████▏ | Scoring GeneralizingEstimator : 756/1225 [00:02<00:01, 299.03it/s]
63%|██████▎ | Scoring GeneralizingEstimator : 768/1225 [00:02<00:01, 301.77it/s]
64%|██████▎ | Scoring GeneralizingEstimator : 780/1225 [00:03<00:01, 304.06it/s]
65%|██████▍ | Scoring GeneralizingEstimator : 792/1225 [00:03<00:01, 306.29it/s]
66%|██████▌ | Scoring GeneralizingEstimator : 803/1225 [00:03<00:01, 307.22it/s]
67%|██████▋ | Scoring GeneralizingEstimator : 815/1225 [00:03<00:01, 309.59it/s]
68%|██████▊ | Scoring GeneralizingEstimator : 827/1225 [00:03<00:01, 311.68it/s]
68%|██████▊ | Scoring GeneralizingEstimator : 839/1225 [00:03<00:01, 313.78it/s]
69%|██████▉ | Scoring GeneralizingEstimator : 850/1225 [00:03<00:01, 314.36it/s]
70%|███████ | Scoring GeneralizingEstimator : 862/1225 [00:03<00:01, 312.97it/s]
71%|███████ | Scoring GeneralizingEstimator : 872/1225 [00:03<00:01, 312.10it/s]
72%|███████▏ | Scoring GeneralizingEstimator : 883/1225 [00:03<00:01, 312.51it/s]
73%|███████▎ | Scoring GeneralizingEstimator : 896/1225 [00:03<00:01, 315.89it/s]
74%|███████▍ | Scoring GeneralizingEstimator : 908/1225 [00:03<00:00, 317.59it/s]
75%|███████▌ | Scoring GeneralizingEstimator : 921/1225 [00:03<00:00, 320.79it/s]
76%|███████▌ | Scoring GeneralizingEstimator : 933/1225 [00:03<00:00, 322.38it/s]
77%|███████▋ | Scoring GeneralizingEstimator : 944/1225 [00:03<00:00, 322.19it/s]
78%|███████▊ | Scoring GeneralizingEstimator : 955/1225 [00:03<00:00, 322.17it/s]
79%|███████▉ | Scoring GeneralizingEstimator : 965/1225 [00:03<00:00, 320.67it/s]
80%|███████▉ | Scoring GeneralizingEstimator : 976/1225 [00:03<00:00, 320.84it/s]
81%|████████ | Scoring GeneralizingEstimator : 988/1225 [00:03<00:00, 322.42it/s]
82%|████████▏ | Scoring GeneralizingEstimator : 1000/1225 [00:03<00:00, 323.72it/s]
83%|████████▎ | Scoring GeneralizingEstimator : 1011/1225 [00:03<00:00, 323.37it/s]
83%|████████▎ | Scoring GeneralizingEstimator : 1020/1225 [00:03<00:00, 320.57it/s]
84%|████████▍ | Scoring GeneralizingEstimator : 1032/1225 [00:03<00:00, 322.27it/s]
85%|████████▌ | Scoring GeneralizingEstimator : 1043/1225 [00:03<00:00, 322.37it/s]
86%|████████▌ | Scoring GeneralizingEstimator : 1053/1225 [00:03<00:00, 321.04it/s]
87%|████████▋ | Scoring GeneralizingEstimator : 1063/1225 [00:03<00:00, 319.52it/s]
88%|████████▊ | Scoring GeneralizingEstimator : 1073/1225 [00:03<00:00, 318.35it/s]
88%|████████▊ | Scoring GeneralizingEstimator : 1083/1225 [00:03<00:00, 317.19it/s]
89%|████████▉ | Scoring GeneralizingEstimator : 1092/1225 [00:03<00:00, 314.62it/s]
90%|████████▉ | Scoring GeneralizingEstimator : 1102/1225 [00:04<00:00, 312.71it/s]
91%|█████████ | Scoring GeneralizingEstimator : 1114/1225 [00:04<00:00, 314.79it/s]
92%|█████████▏| Scoring GeneralizingEstimator : 1125/1225 [00:04<00:00, 315.03it/s]
93%|█████████▎| Scoring GeneralizingEstimator : 1135/1225 [00:04<00:00, 314.08it/s]
93%|█████████▎| Scoring GeneralizingEstimator : 1144/1225 [00:04<00:00, 311.29it/s]
94%|█████████▍| Scoring GeneralizingEstimator : 1153/1225 [00:04<00:00, 308.80it/s]
95%|█████████▌| Scoring GeneralizingEstimator : 1164/1225 [00:04<00:00, 309.56it/s]
96%|█████████▌| Scoring GeneralizingEstimator : 1177/1225 [00:04<00:00, 313.27it/s]
97%|█████████▋| Scoring GeneralizingEstimator : 1188/1225 [00:04<00:00, 313.83it/s]
98%|█████████▊| Scoring GeneralizingEstimator : 1197/1225 [00:04<00:00, 309.99it/s]
99%|█████████▊| Scoring GeneralizingEstimator : 1208/1225 [00:04<00:00, 310.35it/s]
100%|█████████▉| Scoring GeneralizingEstimator : 1220/1225 [00:04<00:00, 312.32it/s]
100%|██████████| Scoring GeneralizingEstimator : 1225/1225 [00:04<00:00, 276.81it/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()
References#
Total running time of the script: (0 minutes 9.969 seconds)
Estimated memory usage: 129 MB