Note
Click here 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
import matplotlib.pyplot as plt
from sklearn.pipeline import make_pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import LogisticRegression
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., 30., 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 sec)
[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.
[Parallel(n_jobs=1)]: Done 1 out of 1 | elapsed: 0.0s remaining: 0.0s
[Parallel(n_jobs=1)]: Done 2 out of 2 | elapsed: 0.0s remaining: 0.0s
[Parallel(n_jobs=1)]: Done 3 out of 3 | elapsed: 0.0s remaining: 0.0s
[Parallel(n_jobs=1)]: Done 4 out of 4 | elapsed: 0.0s remaining: 0.0s
[Parallel(n_jobs=1)]: Done 366 out of 366 | elapsed: 0.7s finished
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(),
y=epochs['Left'].events[:, 2] > 2)
0%| | Fitting GeneralizingEstimator : 0/35 [00:00<?, ?it/s]
6%|5 | Fitting GeneralizingEstimator : 2/35 [00:00<00:00, 58.42it/s]
11%|#1 | Fitting GeneralizingEstimator : 4/35 [00:00<00:00, 58.79it/s]
17%|#7 | Fitting GeneralizingEstimator : 6/35 [00:00<00:00, 58.93it/s]
23%|##2 | Fitting GeneralizingEstimator : 8/35 [00:00<00:00, 58.94it/s]
29%|##8 | Fitting GeneralizingEstimator : 10/35 [00:00<00:00, 58.94it/s]
37%|###7 | Fitting GeneralizingEstimator : 13/35 [00:00<00:00, 64.51it/s]
46%|####5 | Fitting GeneralizingEstimator : 16/35 [00:00<00:00, 68.47it/s]
51%|#####1 | Fitting GeneralizingEstimator : 18/35 [00:00<00:00, 67.09it/s]
60%|###### | Fitting GeneralizingEstimator : 21/35 [00:00<00:00, 69.84it/s]
66%|######5 | Fitting GeneralizingEstimator : 23/35 [00:00<00:00, 68.50it/s]
71%|#######1 | Fitting GeneralizingEstimator : 25/35 [00:00<00:00, 67.41it/s]
74%|#######4 | Fitting GeneralizingEstimator : 26/35 [00:00<00:00, 63.28it/s]
83%|########2 | Fitting GeneralizingEstimator : 29/35 [00:00<00:00, 65.87it/s]
89%|########8 | Fitting GeneralizingEstimator : 31/35 [00:00<00:00, 65.21it/s]
97%|#########7| Fitting GeneralizingEstimator : 34/35 [00:00<00:00, 67.38it/s]
100%|##########| Fitting GeneralizingEstimator : 35/35 [00:00<00:00, 67.29it/s]
Score on the epochs where the stimulus was presented to the right.
scores = time_gen.score(X=epochs['Right'].get_data(),
y=epochs['Right'].events[:, 2] > 2)
0%| | Scoring GeneralizingEstimator : 0/1225 [00:00<?, ?it/s]
1%| | Scoring GeneralizingEstimator : 11/1225 [00:00<00:04, 272.10it/s]
2%|1 | Scoring GeneralizingEstimator : 21/1225 [00:00<00:04, 282.54it/s]
3%|2 | Scoring GeneralizingEstimator : 36/1225 [00:00<00:03, 335.80it/s]
4%|3 | Scoring GeneralizingEstimator : 48/1225 [00:00<00:03, 340.25it/s]
5%|4 | Scoring GeneralizingEstimator : 59/1225 [00:00<00:03, 336.89it/s]
6%|5 | Scoring GeneralizingEstimator : 72/1225 [00:00<00:03, 345.56it/s]
7%|6 | Scoring GeneralizingEstimator : 85/1225 [00:00<00:03, 351.78it/s]
8%|8 | Scoring GeneralizingEstimator : 98/1225 [00:00<00:03, 356.65it/s]
9%|9 | Scoring GeneralizingEstimator : 111/1225 [00:00<00:03, 360.10it/s]
10%|# | Scoring GeneralizingEstimator : 125/1225 [00:00<00:03, 366.52it/s]
11%|#1 | Scoring GeneralizingEstimator : 138/1225 [00:00<00:02, 368.38it/s]
12%|#2 | Scoring GeneralizingEstimator : 152/1225 [00:00<00:02, 373.32it/s]
13%|#3 | Scoring GeneralizingEstimator : 165/1225 [00:00<00:02, 374.42it/s]
15%|#4 | Scoring GeneralizingEstimator : 178/1225 [00:00<00:02, 375.28it/s]
16%|#5 | Scoring GeneralizingEstimator : 190/1225 [00:00<00:02, 373.39it/s]
16%|#6 | Scoring GeneralizingEstimator : 201/1225 [00:00<00:02, 369.06it/s]
17%|#7 | Scoring GeneralizingEstimator : 213/1225 [00:00<00:02, 367.77it/s]
18%|#8 | Scoring GeneralizingEstimator : 225/1225 [00:00<00:02, 366.43it/s]
19%|#9 | Scoring GeneralizingEstimator : 237/1225 [00:00<00:02, 365.54it/s]
20%|## | Scoring GeneralizingEstimator : 248/1225 [00:00<00:02, 362.37it/s]
21%|##1 | Scoring GeneralizingEstimator : 259/1225 [00:00<00:02, 356.75it/s]
22%|##1 | Scoring GeneralizingEstimator : 269/1225 [00:00<00:02, 352.17it/s]
23%|##3 | Scoring GeneralizingEstimator : 283/1225 [00:00<00:02, 356.51it/s]
24%|##4 | Scoring GeneralizingEstimator : 296/1225 [00:00<00:02, 358.38it/s]
25%|##5 | Scoring GeneralizingEstimator : 308/1225 [00:00<00:02, 358.05it/s]
26%|##6 | Scoring GeneralizingEstimator : 324/1225 [00:00<00:02, 365.59it/s]
28%|##7 | Scoring GeneralizingEstimator : 339/1225 [00:00<00:02, 370.67it/s]
29%|##8 | Scoring GeneralizingEstimator : 352/1225 [00:00<00:02, 371.22it/s]
30%|##9 | Scoring GeneralizingEstimator : 367/1225 [00:00<00:02, 375.79it/s]
31%|###1 | Scoring GeneralizingEstimator : 383/1225 [00:01<00:02, 381.74it/s]
33%|###2 | Scoring GeneralizingEstimator : 401/1225 [00:01<00:02, 391.15it/s]
34%|###4 | Scoring GeneralizingEstimator : 420/1225 [00:01<00:02, 401.70it/s]
36%|###5 | Scoring GeneralizingEstimator : 439/1225 [00:01<00:01, 411.48it/s]
37%|###7 | Scoring GeneralizingEstimator : 456/1225 [00:01<00:01, 416.97it/s]
39%|###8 | Scoring GeneralizingEstimator : 472/1225 [00:01<00:01, 420.36it/s]
40%|#### | Scoring GeneralizingEstimator : 490/1225 [00:01<00:01, 426.96it/s]
41%|####1 | Scoring GeneralizingEstimator : 506/1225 [00:01<00:01, 429.52it/s]
43%|####2 | Scoring GeneralizingEstimator : 523/1225 [00:01<00:01, 433.76it/s]
44%|####4 | Scoring GeneralizingEstimator : 539/1225 [00:01<00:01, 435.99it/s]
45%|####5 | Scoring GeneralizingEstimator : 553/1225 [00:01<00:01, 434.68it/s]
46%|####6 | Scoring GeneralizingEstimator : 567/1225 [00:01<00:01, 433.51it/s]
47%|####7 | Scoring GeneralizingEstimator : 581/1225 [00:01<00:01, 432.28it/s]
48%|####8 | Scoring GeneralizingEstimator : 593/1225 [00:01<00:01, 427.88it/s]
49%|####9 | Scoring GeneralizingEstimator : 606/1225 [00:01<00:01, 425.39it/s]
51%|##### | Scoring GeneralizingEstimator : 622/1225 [00:01<00:01, 428.01it/s]
52%|#####1 | Scoring GeneralizingEstimator : 636/1225 [00:01<00:01, 427.21it/s]
53%|#####3 | Scoring GeneralizingEstimator : 650/1225 [00:01<00:01, 426.44it/s]
54%|#####4 | Scoring GeneralizingEstimator : 664/1225 [00:01<00:01, 425.79it/s]
55%|#####5 | Scoring GeneralizingEstimator : 679/1225 [00:01<00:01, 426.73it/s]
57%|#####6 | Scoring GeneralizingEstimator : 695/1225 [00:01<00:01, 429.28it/s]
58%|#####7 | Scoring GeneralizingEstimator : 710/1225 [00:01<00:01, 429.94it/s]
59%|#####9 | Scoring GeneralizingEstimator : 725/1225 [00:01<00:01, 430.62it/s]
60%|###### | Scoring GeneralizingEstimator : 740/1225 [00:01<00:01, 431.11it/s]
62%|######1 | Scoring GeneralizingEstimator : 756/1225 [00:01<00:01, 433.28it/s]
63%|######2 | Scoring GeneralizingEstimator : 770/1225 [00:01<00:01, 432.17it/s]
64%|######3 | Scoring GeneralizingEstimator : 783/1225 [00:01<00:01, 428.86it/s]
65%|######5 | Scoring GeneralizingEstimator : 797/1225 [00:01<00:00, 428.11it/s]
66%|######6 | Scoring GeneralizingEstimator : 811/1225 [00:01<00:00, 427.41it/s]
67%|######7 | Scoring GeneralizingEstimator : 823/1225 [00:02<00:00, 423.56it/s]
69%|######8 | Scoring GeneralizingEstimator : 840/1225 [00:02<00:00, 427.67it/s]
70%|####### | Scoring GeneralizingEstimator : 859/1225 [00:02<00:00, 434.60it/s]
72%|#######1 | Scoring GeneralizingEstimator : 879/1225 [00:02<00:00, 442.70it/s]
73%|#######3 | Scoring GeneralizingEstimator : 898/1225 [00:02<00:00, 448.81it/s]
75%|#######4 | Scoring GeneralizingEstimator : 918/1225 [00:02<00:00, 456.21it/s]
76%|#######6 | Scoring GeneralizingEstimator : 937/1225 [00:02<00:00, 461.56it/s]
78%|#######8 | Scoring GeneralizingEstimator : 958/1225 [00:02<00:00, 469.79it/s]
80%|#######9 | Scoring GeneralizingEstimator : 977/1225 [00:02<00:00, 474.45it/s]
81%|########1 | Scoring GeneralizingEstimator : 994/1225 [00:02<00:00, 475.74it/s]
82%|########2 | Scoring GeneralizingEstimator : 1009/1225 [00:02<00:00, 473.83it/s]
84%|########3 | Scoring GeneralizingEstimator : 1023/1225 [00:02<00:00, 470.42it/s]
85%|########4 | Scoring GeneralizingEstimator : 1037/1225 [00:02<00:00, 467.43it/s]
86%|########5 | Scoring GeneralizingEstimator : 1050/1225 [00:02<00:00, 462.65it/s]
87%|########6 | Scoring GeneralizingEstimator : 1062/1225 [00:02<00:00, 457.17it/s]
88%|########7 | Scoring GeneralizingEstimator : 1073/1225 [00:02<00:00, 450.35it/s]
89%|########8 | Scoring GeneralizingEstimator : 1085/1225 [00:02<00:00, 445.38it/s]
90%|########9 | Scoring GeneralizingEstimator : 1100/1225 [00:02<00:00, 445.31it/s]
91%|######### | Scoring GeneralizingEstimator : 1113/1225 [00:02<00:00, 442.12it/s]
92%|#########2| Scoring GeneralizingEstimator : 1131/1225 [00:02<00:00, 446.64it/s]
94%|#########3| Scoring GeneralizingEstimator : 1149/1225 [00:02<00:00, 450.90it/s]
95%|#########5| Scoring GeneralizingEstimator : 1164/1225 [00:02<00:00, 450.41it/s]
96%|#########6| Scoring GeneralizingEstimator : 1179/1225 [00:02<00:00, 450.08it/s]
98%|#########7| Scoring GeneralizingEstimator : 1198/1225 [00:02<00:00, 455.75it/s]
99%|#########9| Scoring GeneralizingEstimator : 1217/1225 [00:02<00:00, 461.17it/s]
100%|##########| Scoring GeneralizingEstimator : 1225/1225 [00:02<00:00, 431.44it/s]
Plot
fig, ax = plt.subplots(1)
im = ax.matshow(scores, vmin=0, vmax=1., cmap='RdBu_r', origin='lower',
extent=epochs.times[[0, -1, 0, -1]])
ax.axhline(0., color='k')
ax.axvline(0., color='k')
ax.xaxis.set_ticks_position('bottom')
ax.set_xlabel('Testing Time (s)')
ax.set_ylabel('Training Time (s)')
ax.set_title('Generalization across time and condition')
plt.colorbar(im, ax=ax)
plt.show()
References#
- 1
Jean-Rémi King and Stanislas Dehaene. Characterizing the dynamics of mental representations: the temporal generalization method. Trends in Cognitive Sciences, 18(4):203–210, 2014. doi:10.1016/j.tics.2014.01.002.
Total running time of the script: ( 0 minutes 8.383 seconds)
Estimated memory usage: 129 MB