def get_correlations(config):
logdir = os.path.join(config.get('paths', 'log_dir_of_current_run'),
'log_vars', '0.npz')
logvars = np.load(logdir, allow_pickle=True)
spiketrains = logvars['spiketrains_n_b_l_t']
activations = logvars['activations_n_b_l']
spikerates = spiketrains_to_rates(spiketrains,
config.getint('simulation', 'duration'),
config.get('conversion', 'spike_code'))
max_rate = 1. / config.getfloat('simulation', 'dt')
co = get_pearson_coefficients(spikerates, activations, max_rate)
return np.mean(co, axis=1)
def plot_pearson_coefficients(spikerates_batch,
activations_batch,
config,
path=None):
"""
Plot the Pearson correlation coefficients for each layer, averaged over one
mini batch.
Parameters
----------
spikerates_batch: list[tuple[np.array, str]]
Each entry in ``spikerates_batch`` contains a tuple
``(spikerates, label)`` for each layer of the network (for the first
batch only, and excluding ``Flatten`` layers).
``spikerates`` contains the average firing rates of all neurons in a
layer. It has the same shape as the original layer, e.g.
(batch_size, n_features, n_rows, n_cols) for a convolution layer.
``label`` is a string specifying both the layer type and the index,
e.g. ``'03Dense'``.
activations_batch: list[tuple[np.array, str]]
Contains the activations of a net. Same structure as
``spikerates_batch``.
config: configparser.ConfigParser
Settings.
path: Optional[str]
Where to save the output.
"""
from snntoolbox.utils.utils import extract_label
max_rate = 1. / config.getfloat('simulation', 'dt')
co = get_pearson_coefficients(spikerates_batch, activations_batch,
max_rate)
# Average over batch
corr = np.mean(co, axis=1)
std = np.std(co, axis=1)
labels = [sp[1] for sp in spikerates_batch]
if config.getboolean('output', 'use_simple_labels'):
labels = [extract_label(label)[1] for label in labels]
plt.figure()
plt.bar([i + 0.1 for i in range(len(corr))],
corr,
width=0.8,
yerr=std,
color=(0.8, 0.8, 0.8))
plt.ylim([0, 1])
plt.xlim([0, len(corr)])
plt.xticks([i + 0.5 for i in range(len(labels))], labels, rotation=90)
plt.tick_params(bottom=False)
plt.title('Correlation between ANN activations \n and SNN spikerates,\n ' +
'averaged over {} samples'.format(len(spikerates_batch[0][0])))
plt.ylabel('Pearson Correlation Coefficient')
if path is not None:
filename = 'Pearson'
plt.savefig(os.path.join(path, filename), bbox_inches='tight')
else:
plt.show()
plt.close()