def test_add(self):
t1 = make_sparse_labeled_tensor(2)
t2 = make_sparse_labeled_tensor(2)
t1['cat', 'mammal'] = 1
t1['cat', 'pet'] = 1
t1['panda', 'mammal'] = 1
t2['cat', 'pet'] = 1
t2['bear', 'pet'] = -1
t3 = t1 + t2
def test_add(self):
t1 = make_sparse_labeled_tensor(2)
t2 = make_sparse_labeled_tensor(2)
t1['cat', 'mammal'] = 1
t1['cat', 'pet'] = 1
t1['panda', 'mammal'] = 1
t2['cat', 'pet'] = 1
t2['bear', 'pet'] = -1
t3 = t1 + t2
def iteration(self, u, learn=False):
"""Train the eigenvector table with new vector u"""
print "iteration = ", self._iteration
mags = []
u_copy = make_sparse_labeled_tensor(1, initial=u) # copy
for k in xrange(min(self._k, self._iteration+1)):
mag, new_u = self.update_eigenvector(k, u_copy, learn)
u_copy = new_u
mags.append(mag)
if learn:
self._iteration += 1
self._v[1:] = sorted(self._v[1:], reverse=True, key=lambda v: v.norm())
return mags
def iteration(self, u, learn=False):
"""Train the eigenvector table with new vector u"""
print "iteration = ", self._iteration
mags = []
u_copy = make_sparse_labeled_tensor(1, initial=u) # copy
for k in xrange(min(self._k, self._iteration + 1)):
mag, new_u = self.update_eigenvector(k, u_copy, learn)
u_copy = new_u
mags.append(mag)
if learn:
self._iteration += 1
self._v[1:] = sorted(self._v[1:],
reverse=True,
key=lambda v: v.norm())
return mags
def update_eigenvector(self, k, u, learn=False):
"""Update eigenvector k with vector u, returning pair
containing magnitude of eigenvector component and residue vector"""
if learn:
# Handle elementary cases
if self._iteration < k:
return 0.0, zerovec()
if self._iteration == k:
self._v[k] = make_sparse_labeled_tensor(1, initial=u)
mag = self._v[k].norm()
return mag, zerovec()
# Compute weighting factors
n = min(self._iteration, self._remembrance)
if n < self._bootstrap:
w_old = float(n - 1) / n
w_new = 1.0 / n
else:
l = self._amnesia
w_old = float(n - l) / n
w_new = float(l) / n
# Compute the attractor
attractor = self.compute_attractor(k, u)
# Approach attractor
self._v[k] *= w_old
self._v[k] += attractor * w_new
# Calculate component magnitudes
v_hat = self._v[k].hat()
if k == 0:
if self._auto_baseline: base_mag = self._v[k].norm()
else: base_mag = 0.0
else: base_mag = u * v_hat
u_residue = u - (v_hat * base_mag)
if k == 0:
logger.debug('e0: %s' % str(self._v[k]))
logger.debug('u: %s' % str(u))
if learn:
logger.debug('attractor: %s' % str(attractor))
logger.debug('residue: %s' % str(u_residue))
logger.debug('')
return base_mag, u_residue
def make_twit_vec(text, extras=None):
words = list(clean_twitter(text)) or ['empty']
twitvec = make_sparse_labeled_tensor(ndim=1)
for word in words:
if word.strip():
if word[0] == '-':
twitvec[word[1:]] += -1
else:
twitvec[word] += 1
twitvec = twitvec.hat()
if extras is not None:
for extra in extras.split():
if extra[0] == '-':
twitvec[extra[1:]] = -0.1
else:
twitvec[extra] = 0.1
return twitvec
def make_twit_vec(text, extras=None):
words = list(clean_twitter(text)) or [ 'empty' ]
twitvec = make_sparse_labeled_tensor(ndim=1)
for word in words:
if word.strip():
if word[0] == '-':
twitvec[word[1:]] += -1
else:
twitvec[word] += 1
twitvec = twitvec.hat()
if extras is not None:
for extra in extras.split():
if extra[0] == '-':
twitvec[extra[1:]] = -0.1
else:
twitvec[extra] = 0.1
return twitvec
def update_eigenvector(self, k, u, learn=False):
"""Update eigenvector k with vector u, returning pair
containing magnitude of eigenvector component and residue vector"""
if learn:
# Handle elementary cases
if self._iteration < k:
return 0.0, zerovec()
if self._iteration == k:
self._v[k] = make_sparse_labeled_tensor(1, initial=u)
mag = self._v[k].norm()
return mag, zerovec()
# Compute weighting factors
n = min(self._iteration, self._remembrance)
if n < self._bootstrap:
w_old = float(n - 1) / n
w_new = 1.0 / n
else:
l = self._amnesia
w_old = float(n - l) / n
w_new = float(l) / n
# Compute the attractor
attractor = self.compute_attractor(k, u)
# Approach attractor
self._v[k] *= w_old
self._v[k] += attractor * w_new
# Calculate component magnitudes
v_hat = self._v[k].hat()
if k == 0:
if self._auto_baseline: base_mag = self._v[k].norm()
else: base_mag = 0.0
else: base_mag = u * v_hat
u_residue = u - (v_hat * base_mag)
if k == 0:
logger.debug('e0: %s' % str(self._v[k]))
logger.debug('u: %s' % str(u))
if learn:
logger.debug('attractor: %s' % str(attractor))
logger.debug('residue: %s' % str(u_residue))
logger.debug('')
return base_mag, u_residue
def test_make_shape():
labels = OrderedSet(list("abcde"))
t = make_sparse_labeled_tensor(ndim=1, labels=[labels])
eq_(t.shape[0], len(labels))
eq_(t.tensor.shape[0], len(labels))
def ez_matrix(rows, cols, vals):
'''
Simple way to make a matrix (2D sparse tensor).
'''
return make_sparse_labeled_tensor(ndim=2, initial=zip(zip(rows, cols), vals))
def zerovec():
return make_sparse_labeled_tensor(1)
def zerovec():
return make_sparse_labeled_tensor(1)
def test_make_shape():
labels = OrderedSet(list('abcde'))
t = make_sparse_labeled_tensor(ndim=1, labels=[labels])
eq_(t.shape[0], len(labels))
eq_(t.tensor.shape[0], len(labels))
def setUp(self):
self.tensor = make_sparse_labeled_tensor(2)
def ez_matrix(rows, cols, vals):
'''
Simple way to make a matrix (2D sparse tensor).
'''
return make_sparse_labeled_tensor(ndim=2,
initial=zip(zip(rows, cols), vals))
def stack_contains(self, cls):
from csc.divisi.labeled_view import make_sparse_labeled_tensor
return make_sparse_labeled_tensor(2).stack_contains(cls)
def test_wrong_dims():
from csc.divisi.labeled_view import make_sparse_labeled_tensor
t1 = make_sparse_labeled_tensor(ndim=1)
t2 = make_sparse_labeled_tensor(ndim=2)
Blend([t1, t2])
def stack_contains(self, cls):
from csc.divisi.labeled_view import make_sparse_labeled_tensor
return make_sparse_labeled_tensor(2).stack_contains(cls)
def test_wrong_dims():
from csc.divisi.labeled_view import make_sparse_labeled_tensor
t1 = make_sparse_labeled_tensor(ndim=1)
t2 = make_sparse_labeled_tensor(ndim=2)
Blend([t1, t2])
def setUp(self):
self.tensor = make_sparse_labeled_tensor(2)
