braindecode 0.1.6.2__tar.gz → 0.1.6.4__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/Braindecode.egg-info/PKG-INFO +1 -1
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/Braindecode.egg-info/SOURCES.txt +4 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/PKG-INFO +1 -1
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/models/eegnet.py +9 -6
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/models/shallow_fbcsp.py +3 -3
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/torch_ext/util.py +3 -0
- Braindecode-0.1.6.4/braindecode/visualization/__init__.py +0 -0
- Braindecode-0.1.6.4/braindecode/visualization/perturbation.py +92 -0
- Braindecode-0.1.6.4/braindecode/visualization/plot.py +175 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/docs/source/braindecode.rst +1 -0
- Braindecode-0.1.6.4/docs/source/braindecode.visualization.rst +28 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/setup.py +1 -1
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/Braindecode.egg-info/dependency_links.txt +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/Braindecode.egg-info/not-zip-safe +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/Braindecode.egg-info/requires.txt +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/Braindecode.egg-info/top_level.txt +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/LICENSE.txt +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/MANIFEST.in +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/README.rst +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/__init__.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/datasets/__init__.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/datasets/bbci.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/datasets/bcic_iv_2a.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/datasets/multiple.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/datasets/sensor_positions.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/datautil/__init__.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/datautil/iterators.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/datautil/signal_target.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/datautil/signalproc.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/datautil/splitters.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/datautil/trial_segment.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/datautil/util.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/experiments/__init__.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/experiments/experiment.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/experiments/monitors.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/experiments/stopcriteria.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/mne_ext/__init__.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/mne_ext/signalproc.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/models/__init__.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/models/deep4.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/models/util.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/torch_ext/__init__.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/torch_ext/constraints.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/torch_ext/functions.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/torch_ext/init.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/torch_ext/losses.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/torch_ext/modules.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/braindecode/util.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/docs/Makefile +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/docs/conf.py +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/docs/index.rst +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/docs/notebooks/BBCI_Data.ipynb +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/docs/notebooks/BBCI_Data_Start_Stop.ipynb +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/docs/notebooks/BBCI_Data_Trialwise.ipynb +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/docs/notebooks/Cropped_Decoding.ipynb +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/docs/notebooks/Experiment_Class.ipynb +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/docs/notebooks/TrialWise_Decoding.ipynb +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/docs/notebooks/visualization/Perturbation.ipynb +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/docs/source/braindecode.datasets.rst +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/docs/source/braindecode.datautil.rst +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/docs/source/braindecode.experiments.rst +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/docs/source/braindecode.mne_ext.rst +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/docs/source/braindecode.models.rst +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/docs/source/braindecode.torch_ext.rst +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/docs/source/modules.rst +0 -0
- {Braindecode-0.1.6.2 → Braindecode-0.1.6.4}/setup.cfg +0 -0
|
@@ -40,6 +40,9 @@ braindecode/torch_ext/init.py
|
|
|
40
40
|
braindecode/torch_ext/losses.py
|
|
41
41
|
braindecode/torch_ext/modules.py
|
|
42
42
|
braindecode/torch_ext/util.py
|
|
43
|
+
braindecode/visualization/__init__.py
|
|
44
|
+
braindecode/visualization/perturbation.py
|
|
45
|
+
braindecode/visualization/plot.py
|
|
43
46
|
docs/Makefile
|
|
44
47
|
docs/conf.py
|
|
45
48
|
docs/index.rst
|
|
@@ -57,4 +60,5 @@ docs/source/braindecode.mne_ext.rst
|
|
|
57
60
|
docs/source/braindecode.models.rst
|
|
58
61
|
docs/source/braindecode.rst
|
|
59
62
|
docs/source/braindecode.torch_ext.rst
|
|
63
|
+
docs/source/braindecode.visualization.rst
|
|
60
64
|
docs/source/modules.rst
|
|
@@ -48,7 +48,7 @@ class EEGNet(object):
|
|
|
48
48
|
model.add_module('conv_1', nn.Conv2d(
|
|
49
49
|
self.in_chans, n_filters_1, (1, 1), stride=1, bias=True))
|
|
50
50
|
model.add_module('bnorm_1', nn.BatchNorm2d(
|
|
51
|
-
n_filters_1, momentum=0.
|
|
51
|
+
n_filters_1, momentum=0.01, affine=True, eps=1e-3),)
|
|
52
52
|
model.add_module('elu_1', Expression(elu))
|
|
53
53
|
# transpose to examples x 1 x (virtual, not EEG) channels x time
|
|
54
54
|
model.add_module('permute_1', Expression(lambda x: x.permute(0,3,1,2)))
|
|
@@ -57,12 +57,15 @@ class EEGNet(object):
|
|
|
57
57
|
|
|
58
58
|
n_filters_2 = 4
|
|
59
59
|
# not clear to me how they did the padding
|
|
60
|
+
# keras padds unequal padding more in front:
|
|
61
|
+
# https://stackoverflow.com/questions/43994604/padding-with-even-kernel-size-in-a-convolutional-layer-in-keras-theano
|
|
62
|
+
|
|
60
63
|
model.add_module('conv_2', nn.Conv2d(
|
|
61
64
|
1, n_filters_2, self.second_kernel_size, stride=1,
|
|
62
|
-
padding=(self.second_kernel_size[0] // 2,
|
|
65
|
+
padding=(self.second_kernel_size[0] // 2, self.second_kernel_size[1] // 2),
|
|
63
66
|
bias=True))
|
|
64
67
|
model.add_module('bnorm_2',nn.BatchNorm2d(
|
|
65
|
-
n_filters_2, momentum=0.
|
|
68
|
+
n_filters_2, momentum=0.01, affine=True, eps=1e-3),)
|
|
66
69
|
model.add_module('elu_2', Expression(elu))
|
|
67
70
|
model.add_module('pool_2', pool_class(
|
|
68
71
|
kernel_size=(2, 4), stride=(2, 4)))
|
|
@@ -71,10 +74,10 @@ class EEGNet(object):
|
|
|
71
74
|
n_filters_3 = 4
|
|
72
75
|
model.add_module('conv_3', nn.Conv2d(
|
|
73
76
|
n_filters_2, n_filters_3, self.third_kernel_size, stride=1,
|
|
74
|
-
padding=(self.third_kernel_size[0] // 2,
|
|
77
|
+
padding=(self.third_kernel_size[0] // 2, self.third_kernel_size[1] // 2),
|
|
75
78
|
bias=True))
|
|
76
79
|
model.add_module('bnorm_3',nn.BatchNorm2d(
|
|
77
|
-
n_filters_3, momentum=0.
|
|
80
|
+
n_filters_3, momentum=0.01, affine=True, eps=1e-3),)
|
|
78
81
|
model.add_module('elu_3', Expression(elu))
|
|
79
82
|
model.add_module('pool_3', pool_class(
|
|
80
83
|
kernel_size=(2, 4), stride=(2, 4)))
|
|
@@ -108,4 +111,4 @@ class EEGNet(object):
|
|
|
108
111
|
return x
|
|
109
112
|
model.add_module('squeeze', Expression(squeeze_output))
|
|
110
113
|
glorot_weight_zero_bias(model)
|
|
111
|
-
return model
|
|
114
|
+
return model
|
|
@@ -103,9 +103,9 @@ class ShallowFBCSPNet(object):
|
|
|
103
103
|
init.xavier_uniform(model.conv_spat.weight, gain=1)
|
|
104
104
|
if not self.batch_norm:
|
|
105
105
|
init.constant(model.conv_spat.bias, 0)
|
|
106
|
-
|
|
107
|
-
|
|
108
|
-
|
|
106
|
+
if self.batch_norm:
|
|
107
|
+
init.constant(model.bnorm.weight, 1)
|
|
108
|
+
init.constant(model.bnorm.bias, 0)
|
|
109
109
|
init.xavier_uniform(model.conv_classifier.weight, gain=1)
|
|
110
110
|
init.constant(model.conv_classifier.bias, 0)
|
|
111
111
|
|
|
@@ -8,10 +8,13 @@ def np_to_var(X, requires_grad=False, **var_kwargs):
|
|
|
8
8
|
"""
|
|
9
9
|
Convenience function to transform numpy array to `torch.autograd.Variable`.
|
|
10
10
|
|
|
11
|
+
Converts `X` to ndarray using asarray if necessary.
|
|
12
|
+
|
|
11
13
|
Returns
|
|
12
14
|
-------
|
|
13
15
|
var: `torch.autograd.Variable`
|
|
14
16
|
"""
|
|
17
|
+
X = np.asarray(X)
|
|
15
18
|
return Variable(th.from_numpy(X), requires_grad=requires_grad, **var_kwargs)
|
|
16
19
|
|
|
17
20
|
|
|
File without changes
|
|
@@ -0,0 +1,92 @@
|
|
|
1
|
+
import logging
|
|
2
|
+
|
|
3
|
+
import numpy as np
|
|
4
|
+
from numpy.random import RandomState
|
|
5
|
+
|
|
6
|
+
from braindecode.datautil.iterators import get_balanced_batches
|
|
7
|
+
from braindecode.util import wrap_reshape_apply_fn, corr
|
|
8
|
+
|
|
9
|
+
log = logging.getLogger(__name__)
|
|
10
|
+
|
|
11
|
+
|
|
12
|
+
def gaussian_perturbation(amps, rng):
|
|
13
|
+
perturbation = rng.randn(*amps.shape)
|
|
14
|
+
return perturbation
|
|
15
|
+
|
|
16
|
+
|
|
17
|
+
def compute_amplitude_prediction_correlations(pred_fn, examples, n_iterations,
|
|
18
|
+
perturb_fn=gaussian_perturbation,
|
|
19
|
+
batch_size=30,
|
|
20
|
+
seed=((2017, 7, 10))):
|
|
21
|
+
"""
|
|
22
|
+
Perturb input amplitudes and compute correlation between amplitude
|
|
23
|
+
perturbations and prediction changes when pushing perturbed input through
|
|
24
|
+
the prediction function.
|
|
25
|
+
|
|
26
|
+
For more details, see [EEGDeepLearning]_.
|
|
27
|
+
|
|
28
|
+
Parameters
|
|
29
|
+
----------
|
|
30
|
+
pred_fn: function
|
|
31
|
+
Function accepting an numpy input and returning prediction.
|
|
32
|
+
examples: ndarray
|
|
33
|
+
Numpy examples, first axis should be example axis.
|
|
34
|
+
n_iterations: int
|
|
35
|
+
Number of iterations to compute.
|
|
36
|
+
perturb_fn: function, optional
|
|
37
|
+
Function accepting amplitude array and random generator and returning
|
|
38
|
+
perturbation. Default is Gaussian perturbation.
|
|
39
|
+
batch_size: int, optional
|
|
40
|
+
Batch size for computing predictions.
|
|
41
|
+
seed: int, optional
|
|
42
|
+
Random generator seed
|
|
43
|
+
|
|
44
|
+
Returns
|
|
45
|
+
-------
|
|
46
|
+
amplitude_pred_corrs: ndarray
|
|
47
|
+
Correlations between amplitude perturbations and prediction changes
|
|
48
|
+
for all sensors and frequency bins.
|
|
49
|
+
"""
|
|
50
|
+
inds_per_batch = get_balanced_batches(
|
|
51
|
+
n_trials=len(examples), rng=None, shuffle=False, batch_size=batch_size)
|
|
52
|
+
log.info("Compute original predictions...")
|
|
53
|
+
orig_preds = [pred_fn(examples[example_inds])
|
|
54
|
+
for example_inds in inds_per_batch]
|
|
55
|
+
orig_preds_arr = np.concatenate(orig_preds)
|
|
56
|
+
rng = RandomState(seed)
|
|
57
|
+
fft_input = np.fft.rfft(examples, axis=2)
|
|
58
|
+
amps = np.abs(fft_input)
|
|
59
|
+
phases = np.angle(fft_input)
|
|
60
|
+
|
|
61
|
+
amp_pred_corrs = []
|
|
62
|
+
for i_iteration in range(n_iterations):
|
|
63
|
+
log.info("Iteration {:d}...".format(i_iteration))
|
|
64
|
+
log.info("Sample perturbation...")
|
|
65
|
+
perturbation = perturb_fn(amps, rng)
|
|
66
|
+
log.info("Compute new amplitudes...")
|
|
67
|
+
# do not allow perturbation to make amplitudes go below
|
|
68
|
+
# zero
|
|
69
|
+
perturbation = np.maximum(-amps, perturbation)
|
|
70
|
+
new_amps = amps + perturbation
|
|
71
|
+
log.info("Compute new complex inputs...")
|
|
72
|
+
new_complex = amplitude_phase_to_complex(new_amps, phases)
|
|
73
|
+
log.info("Compute new real inputs...")
|
|
74
|
+
new_in = np.fft.irfft(new_complex, axis=2).astype(np.float32)
|
|
75
|
+
log.info("Compute new predictions...")
|
|
76
|
+
new_preds = [pred_fn(new_in[example_inds])
|
|
77
|
+
for example_inds in inds_per_batch]
|
|
78
|
+
|
|
79
|
+
new_preds_arr = np.concatenate(new_preds)
|
|
80
|
+
|
|
81
|
+
diff_preds = new_preds_arr - orig_preds_arr
|
|
82
|
+
|
|
83
|
+
log.info("Compute correlation...")
|
|
84
|
+
amp_pred_corr = wrap_reshape_apply_fn(corr, perturbation[:, :, :, 0],
|
|
85
|
+
diff_preds,
|
|
86
|
+
axis_a=(0,), axis_b=(0))
|
|
87
|
+
amp_pred_corrs.append(amp_pred_corr)
|
|
88
|
+
return amp_pred_corrs
|
|
89
|
+
|
|
90
|
+
|
|
91
|
+
def amplitude_phase_to_complex(amplitude, phase):
|
|
92
|
+
return amplitude * np.cos(phase) + amplitude * np.sin(phase) * 1j
|
|
@@ -0,0 +1,175 @@
|
|
|
1
|
+
# coding=utf-8
|
|
2
|
+
import math
|
|
3
|
+
|
|
4
|
+
import numpy as np
|
|
5
|
+
from scipy import interpolate
|
|
6
|
+
import matplotlib.pyplot as plt
|
|
7
|
+
from matplotlib import patches
|
|
8
|
+
from matplotlib.path import Path
|
|
9
|
+
from matplotlib import cm
|
|
10
|
+
|
|
11
|
+
from braindecode.datasets.sensor_positions import CHANNEL_10_20_APPROX, \
|
|
12
|
+
get_channelpos
|
|
13
|
+
|
|
14
|
+
|
|
15
|
+
def ax_scalp(v, channels,
|
|
16
|
+
ax=None, annotate=False,
|
|
17
|
+
vmin=None, vmax=None, cmap=cm.coolwarm,
|
|
18
|
+
scalp_line_width=1,
|
|
19
|
+
scalp_line_style='solid',
|
|
20
|
+
chan_pos_list=CHANNEL_10_20_APPROX,
|
|
21
|
+
interpolation='bilinear'):
|
|
22
|
+
"""Draw a scalp plot.
|
|
23
|
+
|
|
24
|
+
Draws a scalp plot on an existing axes. The method takes an array of
|
|
25
|
+
values and an array of the corresponding channel names. It matches
|
|
26
|
+
the channel names with an channel position list
|
|
27
|
+
to project them correctly on the scalp.
|
|
28
|
+
|
|
29
|
+
Parameters
|
|
30
|
+
----------
|
|
31
|
+
v : 1d-array of floats
|
|
32
|
+
The values for the channels
|
|
33
|
+
channels : 1d array of strings
|
|
34
|
+
The corresponding channel names for the values in ``v``
|
|
35
|
+
ax : Axes, optional
|
|
36
|
+
The axes to draw the scalp plot on. If not provided, the
|
|
37
|
+
currently activated axes (i.e. ``gca()``) will be taken
|
|
38
|
+
annotate : Boolean, optional
|
|
39
|
+
Draw the channel names next to the channel markers.
|
|
40
|
+
vmin, vmax : float, optional
|
|
41
|
+
The display limits for the values in ``v``. If the data in ``v``
|
|
42
|
+
contains values between -3..3 and ``vmin`` and ``vmax`` are set
|
|
43
|
+
to -1 and 1, all values smaller than -1 and bigger than 1 will
|
|
44
|
+
appear the same as -1 and 1. If not set, the maximum absolute
|
|
45
|
+
value in ``v`` is taken to calculate both values.
|
|
46
|
+
cmap : matplotlib.colors.colormap, optional
|
|
47
|
+
A colormap to define the color transitions.
|
|
48
|
+
scalp_line_width: float
|
|
49
|
+
Line width for outline of scalp
|
|
50
|
+
scalp_line_style: float
|
|
51
|
+
Line style for outline of scalp
|
|
52
|
+
chan_pos_list: iterable of tuples
|
|
53
|
+
First entry should be 'angle' or 'cartesian',
|
|
54
|
+
remaining entries 2-tuples of x and y.
|
|
55
|
+
interpolation: str
|
|
56
|
+
|
|
57
|
+
Returns
|
|
58
|
+
-------
|
|
59
|
+
ax : Axes
|
|
60
|
+
the axes on which the plot was drawn
|
|
61
|
+
|
|
62
|
+
Notes
|
|
63
|
+
-----
|
|
64
|
+
Code adapted from Wyrm [1]_ toolbox https://github.com/bbci/wyrm.
|
|
65
|
+
|
|
66
|
+
References
|
|
67
|
+
----------
|
|
68
|
+
|
|
69
|
+
|
|
70
|
+
.. [1] Schirrmeister, R. T., Springenberg, J. T., Fiederer, L. D. J.,
|
|
71
|
+
Glasstetter, M., Eggensperger, K., Tangermann, M., ... & Ball, T. (2017).
|
|
72
|
+
Deep learning with convolutional neural networks for EEG decoding and
|
|
73
|
+
visualization.
|
|
74
|
+
arXiv preprint arXiv:1703.05051.
|
|
75
|
+
|
|
76
|
+
"""
|
|
77
|
+
if ax is None:
|
|
78
|
+
ax = plt.gca()
|
|
79
|
+
assert len(v) == len(channels), "Should be as many values as channels"
|
|
80
|
+
assert interpolation == 'bilinear' or interpolation == 'nearest'
|
|
81
|
+
if vmin is None:
|
|
82
|
+
# added by me (robintibor@gmail.com)
|
|
83
|
+
assert vmax is None
|
|
84
|
+
vmin, vmax = -np.max(np.abs(v)), np.max(np.abs(v))
|
|
85
|
+
# what if we have an unknown channel?
|
|
86
|
+
points = [get_channelpos(c, chan_pos_list) for c in channels]
|
|
87
|
+
for c in channels:
|
|
88
|
+
assert get_channelpos(c, chan_pos_list) is not None, (
|
|
89
|
+
"Expect " + c + " to exist in positions")
|
|
90
|
+
z = [v[i] for i in range(len(points))]
|
|
91
|
+
# calculate the interpolation
|
|
92
|
+
x = [i[0] for i in points]
|
|
93
|
+
y = [i[1] for i in points]
|
|
94
|
+
# interpolate the in-between values
|
|
95
|
+
xx = np.linspace(min(x), max(x), 500)
|
|
96
|
+
yy = np.linspace(min(y), max(y), 500)
|
|
97
|
+
if interpolation == 'bilinear':
|
|
98
|
+
xx_grid, yy_grid = np.meshgrid(xx, yy)
|
|
99
|
+
f = interpolate.LinearNDInterpolator(list(zip(x, y)), z)
|
|
100
|
+
zz = f(xx_grid, yy_grid)
|
|
101
|
+
else:
|
|
102
|
+
assert interpolation == 'nearest'
|
|
103
|
+
f = interpolate.NearestNDInterpolator(list(zip(x, y)), z)
|
|
104
|
+
assert len(xx) == len(yy)
|
|
105
|
+
zz = np.ones((len(xx), len(yy)))
|
|
106
|
+
for i_x in xrange(len(xx)):
|
|
107
|
+
for i_y in xrange(len(yy)):
|
|
108
|
+
# somehow this is correct. don't know why :(
|
|
109
|
+
zz[i_y, i_x] = f(xx[i_x], yy[i_y])
|
|
110
|
+
# zz[i_x,i_y] = f(xx[i_x], yy[i_y])
|
|
111
|
+
assert not np.any(np.isnan(zz))
|
|
112
|
+
|
|
113
|
+
# plot map
|
|
114
|
+
image = ax.imshow(zz, vmin=vmin, vmax=vmax, cmap=cmap,
|
|
115
|
+
extent=[min(x), max(x), min(y), max(y)], origin='lower',
|
|
116
|
+
interpolation=interpolation)
|
|
117
|
+
# image = ax.contourf(xx, yy, zz, 100, vmin=vmin, vmax=vmax,
|
|
118
|
+
# cmap=colormap)
|
|
119
|
+
if scalp_line_width > 0:
|
|
120
|
+
# paint the head
|
|
121
|
+
ax.add_artist(plt.Circle((0, 0), 1, linestyle=scalp_line_style,
|
|
122
|
+
linewidth=scalp_line_width, fill=False))
|
|
123
|
+
# add a nose
|
|
124
|
+
ax.plot([-0.1, 0, 0.1], [1, 1.1, 1], color='black',
|
|
125
|
+
linewidth=scalp_line_width, linestyle=scalp_line_style)
|
|
126
|
+
# add ears
|
|
127
|
+
_add_ears(ax, scalp_line_width, scalp_line_style)
|
|
128
|
+
# add markers at channels positions
|
|
129
|
+
# set the axes limits, so the figure is centered on the scalp
|
|
130
|
+
ax.set_ylim([-1.05, 1.15])
|
|
131
|
+
ax.set_xlim([-1.15, 1.15])
|
|
132
|
+
|
|
133
|
+
# hide the frame and ticks
|
|
134
|
+
ax.set_frame_on(False)
|
|
135
|
+
ax.set_xticks([])
|
|
136
|
+
ax.set_yticks([])
|
|
137
|
+
# draw the channel names
|
|
138
|
+
if annotate:
|
|
139
|
+
for i in zip(channels, list(zip(x, y))):
|
|
140
|
+
ax.annotate(" " + i[0], i[1], horizontalalignment="center",
|
|
141
|
+
verticalalignment='center')
|
|
142
|
+
ax.set_aspect(1)
|
|
143
|
+
return image
|
|
144
|
+
|
|
145
|
+
|
|
146
|
+
def _add_ears(ax, linewidth, linestyle):
|
|
147
|
+
start_x = np.cos(10 * np.pi / 180.0)
|
|
148
|
+
start_y = np.sin(10 * np.pi / 180.0)
|
|
149
|
+
end_x = np.cos(-15 * np.pi / 180.0)
|
|
150
|
+
end_y = np.sin(-15 * np.pi / 180.0)
|
|
151
|
+
verts = [
|
|
152
|
+
(start_x, start_y),
|
|
153
|
+
(start_x + 0.05, start_y + 0.05), # out up
|
|
154
|
+
(start_x + 0.1, start_y), # further out, back down
|
|
155
|
+
(start_x + 0.11, (end_y * 0.7 + start_y * 0.3)), # midpoint
|
|
156
|
+
(end_x + 0.14, end_y), # down out start
|
|
157
|
+
(end_x + 0.05, end_y - 0.05), # down out further
|
|
158
|
+
(end_x, end_y), # endpoint
|
|
159
|
+
]
|
|
160
|
+
|
|
161
|
+
codes = [Path.MOVETO] + [Path.CURVE3] * (len(verts) - 1)
|
|
162
|
+
|
|
163
|
+
path = Path(verts, codes)
|
|
164
|
+
|
|
165
|
+
patch = patches.PathPatch(path, facecolor='none',
|
|
166
|
+
linestyle=linestyle, linewidth=linewidth)
|
|
167
|
+
|
|
168
|
+
ax.add_patch(patch)
|
|
169
|
+
verts_left = [(-x, y) for x, y in verts]
|
|
170
|
+
path_left = Path(verts_left, codes)
|
|
171
|
+
|
|
172
|
+
patch_left = patches.PathPatch(path_left, facecolor='none',
|
|
173
|
+
linestyle=linestyle, linewidth=linewidth)
|
|
174
|
+
|
|
175
|
+
ax.add_patch(patch_left)
|
|
@@ -0,0 +1,28 @@
|
|
|
1
|
+
braindecode\.visualization package
|
|
2
|
+
==================================
|
|
3
|
+
|
|
4
|
+
.. automodule:: braindecode.visualization
|
|
5
|
+
:members:
|
|
6
|
+
:undoc-members:
|
|
7
|
+
:show-inheritance:
|
|
8
|
+
|
|
9
|
+
Submodules
|
|
10
|
+
----------
|
|
11
|
+
|
|
12
|
+
braindecode\.visualization\.perturbation module
|
|
13
|
+
-----------------------------------------------
|
|
14
|
+
|
|
15
|
+
.. automodule:: braindecode.visualization.perturbation
|
|
16
|
+
:members:
|
|
17
|
+
:undoc-members:
|
|
18
|
+
:show-inheritance:
|
|
19
|
+
|
|
20
|
+
braindecode\.visualization\.plot module
|
|
21
|
+
---------------------------------------
|
|
22
|
+
|
|
23
|
+
.. automodule:: braindecode.visualization.plot
|
|
24
|
+
:members:
|
|
25
|
+
:undoc-members:
|
|
26
|
+
:show-inheritance:
|
|
27
|
+
|
|
28
|
+
|
|
@@ -15,7 +15,7 @@ setup(
|
|
|
15
15
|
# Versions should comply with PEP440. For a discussion on single-sourcing
|
|
16
16
|
# the version across setup.py and the project code, see
|
|
17
17
|
# http://packaging.python.org/en/latest/tutorial.html#version
|
|
18
|
-
version='0.1.6.
|
|
18
|
+
version='0.1.6.4', # TODO: read from __init__.py?
|
|
19
19
|
|
|
20
20
|
description='A deep learning toolbox to decode raw time-domain EEG.',
|
|
21
21
|
long_description=long_description,
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|