def test_out_prediction():
setting = SpatialSettings(debug = False, min_overlap = 1, desired_local_activity = 4, connected_pct = 1,
connected_perm = 0.01, xinput = 3, yinput = 3, potential_radius = 4, xdimension = 3,
ydimension = 3, initial_inhibition_radius = 1, permanence_inc = 0.1, permanence_dec = 0.1,
max_boost = 2, min_duty_cycle_fraction = 0.2)
setting.debug = True
setting.min_overlap = 1
setting.desired_local_activity = 1
setting.connected_pct = 1
setting.xinput = 4
setting.yinput = 4
setting.potential_radius = 2
setting.xdimension = 4
setting.ydimension = 1
setting.initial_inhibition_radius = 2
r = Region(setting, SimpleMapper())
inp = [[1, 0, 1, 0], [1, 0, 1, 0], [1, 0, 1, 0], [1, 0, 1, 0]]
r.step_forward(inp)
res = r.out_prediction([[1, 0, 1, 0]])
# print(res)
for i in range(len(res)):
for j in range(len(res[0])):
if inp[i][j] ==1:
assert res[i][j] > 0
from spatialPooler.sp_region import Region as SpatialPoolerRegion
from apps.settings import *
input_settings = InputSettings(SCALE=1, STEPS_NUMBER=300)
generator = ListImages(12)
spatial_settings = SpatialSettings(debug = False, min_overlap = 1, desired_local_activity = 2, connected_pct = 1,
connected_perm = 0.1, xinput = generator.square_size, yinput = generator.square_size,
potential_radius = 4, xdimension = 12, ydimension = 12, initial_inhibition_radius = 2,
permanence_inc = 0.02, permanence_dec = 0.1, max_boost = 2, min_duty_cycle_fraction = 0.2)
temporal_settings = TemporalSettings(region_size=spatial_settings.xdimension, column_size=4, initial_permanence=0.5,
dendrite_activate_threshold=1, dendrite_permanence_inc_delta=0.02,
dendrite_permanence_dec_delta=-0.1, passive_time_to_active_threshold=1000,
synapse_threshold=0.45)
spatial_pooler_region = SpatialPoolerRegion(spatial_settings=spatial_settings, mapper=SquareMapperAutoRadius())
temporal_pooler_region = TemporalPoolerRegion(temporal_settings=temporal_settings)
for i in range(input_settings.STEPS_NUMBER):
generator.out()
outdata=spatial_pooler_region.step_forward(generator.get_data())
print_matrix(to_binmatrix(outdata))
temporal_pooler_region.step_forward(outdata)
temporal_pooler_region.out_prediction()
cols=temporal_pooler_region.columns
imagined_input = spatial_pooler_region.out_prediction(cols, temporal_settings.region_size)
generator.move()