def testDataMatrix(self):
names = ['a','b','c']
data = [[2,0,0],[0,2,0],[0,0,2]]
obrec = ObjectRecognize()
obrec.label('a', data[0])
obrec.label('b', data[1])
obrec.label('c', data[2])
data = [[2,0,0],[2,2,0],[1,0,2]]
print obrec.getMatchText(names, data)
def testFullFieldStaticImage(self):
print_htm_state = False
h = HTM(cellsPerColumn=1)
h.initialize_input(self.sea_anemone, compressionFactor=3.5)
#learn the different static data images
swap = FlashCards(90, self.sea_anemone, self.seastar,
self.sea_cucumber, self.hermitcrab)
h.execute(swap.dataGenerator(), ticks=90 * 4 * 3 - 1)
#run the htm network through and save the state in a history object
swap = FlashCards(10, self.sea_anemone, self.seastar,
self.sea_cucumber, self.hermitcrab)
history = ExciteHistory(temporal=False)
h.execute(swap.dataGenerator(),
ticks=10 * 4 - 1,
learning=False,
postTick=history.update)
#label the different static data images
recognize = ObjectRecognize()
recognize.label('sea_anemone', history.data[-31])
recognize.label('seastar', history.data[-21])
recognize.label('sea_cucumber', history.data[-11])
recognize.label('hermitcrab', history.data[-1])
if print_htm_state:
print "\n\n*************** Recognition Labeling **************\n\n"
print history.text_graph()
#test recognition of the different static data images
swap.reset()
history = ExciteHistory(temporal=False)
h.execute(swap.dataGenerator(),
ticks=10 * 4 - 1,
learning=False,
postTick=history.update)
if print_htm_state:
print "\n\n*************** Recognition Testing **************\n\n"
print history.text_graph()
#show and test recognition data
testnames = ['sea_anemone', 'seastar', 'sea_cucumber', 'hermitcrab']
testdata = history.data[9:40:10]
print recognize.getMatchText(testnames, testdata)
for x, labelrow in enumerate(
recognize.getMatchData(testnames, testdata)):
for y, match_percent in enumerate(labelrow):
if x == y:
self.assertGreaterEqual(
match_percent,
0.9,
msg=
'htm did not correctly recognize the %s (%.1f%% match)'
% (testnames[x], match_percent * 100))
else:
self.assertLessEqual(
match_percent,
0.8,
msg=
'htm thought it recognized a %s, when it saw a %s (%.1f%% match)'
% (testnames[x], testnames[y], match_percent * 100))
def testFullFieldStaticImage(self):
print_htm_state = False
h = HTM(cellsPerColumn=1)
h.initialize_input(self.sea_anemone)
#learn the different static data images
swap = FlashCards(90, self.sea_anemone, self.seastar, self.sea_cucumber, self.hermitcrab)
h.execute(self.sea_anemone, swap.updateMatrix, ticks=90*4*3)
#run the htm network through and save the state in a history object
swap = FlashCards(10, self.sea_anemone, self.seastar, self.sea_cucumber, self.hermitcrab)
history = ExciteHistory(temporal=False)
h.execute(self.sea_anemone, swap.updateMatrix, ticks=10*4, learning=False, postTick=history.update)
#label the different static data images
recognize = ObjectRecognize()
recognize.label('sea_anemone', history.data[-31])
recognize.label('seastar', history.data[-21])
recognize.label('sea_cucumber', history.data[-11])
recognize.label('hermitcrab', history.data[-1])
if print_htm_state:
print "\n\n*************** Recognition Labeling **************\n\n"
print history.text_graph()
#test recognition of the different static data images
swap.reset()
history = ExciteHistory(temporal=False)
h.execute(self.sea_anemone, swap.updateMatrix, ticks=10*4, learning=False, postTick=history.update)
if print_htm_state:
print "\n\n*************** Recognition Testing **************\n\n"
print history.text_graph()
#show and test recognition data
testnames = ['sea_anemone', 'seastar', 'sea_cucumber', 'hermitcrab']
testdata = history.data[9:40:10]
print recognize.getMatchText(testnames, testdata)
for x, labelrow in enumerate(recognize.getMatchData(testnames, testdata)):
for y, match_percent in enumerate(labelrow):
if x == y:
self.assertGreaterEqual(match_percent, 0.9,
msg='htm did not correctly recognize the %s (%.1f%% match)' % (
testnames[x], match_percent*100))
else:
self.assertLessEqual(match_percent, 0.8,
msg='htm thought it recognized a %s, when it saw a %s (%.1f%% match)' %(
testnames[x], testnames[y], match_percent*100))
