def testUpdatePermanencesForColumn(self):
sp = SpatialPooler(inputDimensions = [5],
columnDimensions = [5])
sp.setSynPermTrimThreshold(0.05)
permanencesList = [
[ -0.10, 0.500, 0.400, 0.010, 0.020 ],
[ 0.300, 0.010, 0.020, 0.120, 0.090 ],
[ 0.070, 0.050, 1.030, 0.190, 0.060 ],
[ 0.180, 0.090, 0.110, 0.010, 0.030 ],
[ 0.200, 0.101, 0.050, -0.09, 1.100 ]]
expectedPermanencesList = [
[ 0.000, 0.500, 0.400, 0.000, 0.000],
# Clip - - Trim Trim
[0.300, 0.000, 0.000, 0.120, 0.090],
# - Trim Trim - -
[0.070, 0.050, 1.000, 0.190, 0.060],
# - - Clip - -
[0.180, 0.090, 0.110, 0.000, 0.000],
# - - - Trim Trim
[0.200, 0.101, 0.050, 0.000, 1.000]]
# - - - Clip Clip
expectedConnectedSynapsesList = [
[0, 1, 1, 0, 0],
[1, 0, 0, 1, 0],
[0, 0, 1, 1, 0],
[1, 0, 1, 0, 0],
[1, 1, 0, 0, 1]]
expectedConnectedCounts = [2, 2, 2, 2, 3]
for i in xrange(5):
permanences = np.array(permanencesList[i], dtype=realDType)
expectedPermanences = np.array(expectedPermanencesList[i],
dtype=realDType)
expectedConnectedSynapses = expectedConnectedSynapsesList[i]
sp._updatePermanencesForColumn(permanences, i, False)
updatedPermanences = np.zeros(5, dtype=realDType)
connectedSynapses = np.zeros(5, dtype=uintDType)
connectedCounts = np.zeros(5, dtype=uintDType)
sp.getPermanence(i, updatedPermanences)
sp.getConnectedSynapses(i, connectedSynapses)
sp.getConnectedCounts(connectedCounts)
np.testing.assert_almost_equal(updatedPermanences, expectedPermanences)
self.assertEqual(list(connectedSynapses), expectedConnectedSynapses)
self.assertEqual(connectedCounts[i], expectedConnectedCounts[i])
def testUpdatePermanencesForColumn(self):
sp = SpatialPooler(inputDimensions=[5], columnDimensions=[5])
sp.setSynPermTrimThreshold(0.05)
permanencesList = [[-0.10, 0.500, 0.400, 0.010, 0.020],
[0.300, 0.010, 0.020, 0.120, 0.090],
[0.070, 0.050, 1.030, 0.190, 0.060],
[0.180, 0.090, 0.110, 0.010, 0.030],
[0.200, 0.101, 0.050, -0.09, 1.100]]
expectedPermanencesList = [
[0.000, 0.500, 0.400, 0.000, 0.000],
# Clip - - Trim Trim
[0.300, 0.000, 0.000, 0.120, 0.090],
# - Trim Trim - -
[0.070, 0.050, 1.000, 0.190, 0.060],
# - - Clip - -
[0.180, 0.090, 0.110, 0.000, 0.000],
# - - - Trim Trim
[0.200, 0.101, 0.050, 0.000, 1.000]
]
# - - - Clip Clip
expectedConnectedSynapsesList = [[0, 1, 1, 0, 0], [1, 0, 0, 1, 0],
[0, 0, 1, 1, 0], [1, 0, 1, 0, 0],
[1, 1, 0, 0, 1]]
expectedConnectedCounts = [2, 2, 2, 2, 3]
for i in xrange(5):
permanences = np.array(permanencesList[i], dtype=realDType)
expectedPermanences = np.array(expectedPermanencesList[i],
dtype=realDType)
expectedConnectedSynapses = expectedConnectedSynapsesList[i]
sp._updatePermanencesForColumn(permanences, i, False)
updatedPermanences = np.zeros(5, dtype=realDType)
connectedSynapses = np.zeros(5, dtype=uintDType)
connectedCounts = np.zeros(5, dtype=uintDType)
sp.getPermanence(i, updatedPermanences)
sp.getConnectedSynapses(i, connectedSynapses)
sp.getConnectedCounts(connectedCounts)
np.testing.assert_almost_equal(updatedPermanences,
expectedPermanences)
self.assertEqual(list(connectedSynapses),
expectedConnectedSynapses)
self.assertEqual(connectedCounts[i], expectedConnectedCounts[i])
