def _create_cfs(self):
"""
Creates the CF objects, initializing the weights one by one
and adding them to the sparse weights object in chunks.
"""
vectorized_create_cf = simple_vectorize(self._create_cf)
self.cfs = vectorized_create_cf(*self._generate_coords())
self.flatcfs = list(self.cfs.flat)
self.weights = sparse.csarray_float(self.src.activity.shape,
self.dest.activity.shape)
cf_x, cf_y = self.dest.activity.shape
src_x, src_y = self.src.activity.shape
y_array = np.zeros((src_x * src_y * cf_y), dtype=np.int32)
x_array = np.zeros((src_x * src_y * cf_y), dtype=np.int32)
val_array = np.zeros((src_x * src_y * cf_y), dtype=np.float32)
# Iterate over the CFs
for x in range(cf_x):
temp_sparse = sparse.csarray_float(self.src.activity.shape,
self.dest.activity.shape)
idx = 0
for y in range(cf_y):
x1, x2, y1, y2 = self.cfs[x][y].input_sheet_slice.tolist()
if self.same_cf_shape_for_all_cfs:
mask_template = self.mask_template
else:
mask_template = _create_mask(self.cf_shape,
self.bounds_template,
self.src, self.autosize_mask,
self.mask_threshold)
weights = self.cfs[x][y]._init_weights(mask_template)
cn_x, cn_y = weights.shape
y_val = x * cf_y + y
for cnx in range(cn_x):
val_array[idx:idx + cn_y] = weights[cnx, :]
x_val = (x1 + cnx) * src_y + y1
x_array[idx:idx + cn_y] = range(x_val, x_val + cn_y)
y_array[idx:idx + cn_y] = y_val
idx += cn_y
nnz_idx = val_array.nonzero()
temp_sparse.setTriplets(x_array[nnz_idx], y_array[nnz_idx],
val_array[nnz_idx])
self.weights += temp_sparse
x_array *= 0
y_array *= 0
val_array *= 0.0
del temp_sparse
self.weights.compress()
self.apply_learn_output_fns()
print self.name, "loaded"
def _create_cfs(self):
"""
Creates the CF objects, initializing the weights one by one
and adding them to the sparse weights object in chunks.
"""
vectorized_create_cf = simple_vectorize(self._create_cf)
self.cfs = vectorized_create_cf(*self._generate_coords())
self.flatcfs = list(self.cfs.flat)
self.weights = sparse.csarray_float(self.src.activity.shape,self.dest.activity.shape)
cf_x,cf_y = self.dest.activity.shape
src_x,src_y = self.src.activity.shape
y_array = np.zeros((src_x*src_y*cf_y),dtype=np.int32)
x_array = np.zeros((src_x*src_y*cf_y),dtype=np.int32)
val_array = np.zeros((src_x*src_y*cf_y),dtype=np.float32)
# Iterate over the CFs
for x in range(cf_x):
temp_sparse = sparse.csarray_float(self.src.activity.shape,self.dest.activity.shape)
idx = 0
for y in range(cf_y):
cf = self.cfs[x][y]
label = cf.label + ('-%d' % self.seed if self.seed is not None else '')
name = "%s_CF (%.5f, %.5f)" % ('' if label is None else label, cf.x,cf.y)
x1,x2,y1,y2 = cf.input_sheet_slice.tolist()
if self.same_cf_shape_for_all_cfs:
mask_template = self.mask_template
else:
mask_template = _create_mask(self.cf_shape,self.bounds_template,
self.src,self.autosize_mask,
self.mask_threshold, name=name)
weights = self.cfs[x][y]._init_weights(mask_template)
cn_x,cn_y = weights.shape
y_val = x * cf_y + y
for cnx in range(cn_x):
val_array[idx:idx+cn_y] = weights[cnx,:]
x_val = (x1+cnx) * src_y + y1
x_array[idx:idx+cn_y] = range(x_val,x_val+cn_y)
y_array[idx:idx+cn_y] = y_val
idx += cn_y
nnz_idx = val_array.nonzero()
temp_sparse.setTriplets(x_array[nnz_idx],y_array[nnz_idx],val_array[nnz_idx])
self.weights += temp_sparse
x_array *= 0; y_array *= 0; val_array *= 0.0
del temp_sparse
self.weights.compress()
self.debug("Sparse projection %r loaded" % self.name)
def __setstate__(self,state_dict):
"""
Method to support unpickling of sparse weights object.
"""
self.__dict__.update(state_dict)
self.weights = sparse.csarray_float(self.weight_shape[0],self.weight_shape[1])
rowInds, colInds, values = self.triplets
self.weights.setTriplets(rowInds,colInds,values)
del self.triplets
del self.weight_shape
def __setstate__(self,state_dict):
"""
Method to support unpickling of sparse weights object.
"""
self.__dict__.update(state_dict)
self.weights = sparse.csarray_float(self.weight_shape[0],self.weight_shape[1])
rowInds, colInds, values = self.triplets
self.weights.setTriplets(rowInds,colInds,values)
del self.triplets
del self.weight_shape
def __call__(self, projection, **params):
time = math.ceil(topo.sim.time())
if self.disk_mask:
self.disk = pattern.Disk(size=1.0,smoothing=0.0)
# Get CF and src sheet shapes
cf_x,cf_y = projection.dest.activity.shape
src_x,src_y = projection.src.activity.shape
# Initialize sparse triplet arrays
y_array = np.zeros((src_x*src_y*cf_y),dtype=np.int32)
x_array = np.zeros((src_x*src_y*cf_y),dtype=np.int32)
val_array = np.zeros((src_x*src_y*cf_y),dtype=sparse_type)
# Create new sparse matrix to accumulate into
sum_sparse = sparse.csarray_float(projection.src.activity.shape,projection.dest.activity.shape)
# Counters for logging
sprout_sum = 0; prune_sum = 0; unit_total = 0
self.mask_total = 0
if (time == 0):
if not hasattr(self,"initial_conns"):
self.initial_conns = {}
self.initial_conns[projection.name] = projection.n_conns()
elif (time % self.interval) == 0:
idx=0
for cidx,cf in enumerate(projection.flatcfs):
temp_weights = cf.weights
dense_unit_mask = (1.0 - (temp_weights>0.0))
dim1,dim2 = temp_weights.shape
sprout_count,prune_idx,nnz = self.calc_ratios(temp_weights)
self.prune(temp_weights,prune_idx)
nnz_pp = np.count_nonzero(temp_weights)
prune_sum += (nnz_pp-nnz)
self.sprout(temp_weights,dense_unit_mask,sprout_count)
nnz_ps = np.count_nonzero(temp_weights)
sprout_sum += nnz_ps - nnz_pp
unit_total += nnz_ps
# Populate sparse array chunk
temp_sparse = sparse.csarray_float(projection.src.activity.shape,projection.dest.activity.shape)
x1,x2,y1,y2 = cf.input_sheet_slice.tolist()
for cnx in range(dim1):
val_array[idx:idx+dim2] = temp_weights[cnx,:]
x_val = (x1+cnx) * src_y + y1
x_array[idx:idx+dim2] = range(x_val,x_val+dim2)
y_array[idx:idx+dim2] = cidx
idx += dim2
# Populate combined sparse array with sparse array chunk
if (cidx+1)%cf_y == 0:
nnz_idx = val_array.nonzero()
temp_sparse.setTriplets(x_array[nnz_idx],y_array[nnz_idx],val_array[nnz_idx])
sum_sparse += temp_sparse
x_array *= 0; y_array *= 0; val_array *= 0.0
idx=0
projection.weights = sum_sparse
del temp_sparse, sum_sparse
projection.weights.compress()
self.message("%s pruned by %d and sprouted %d, connection is now %f%% dense" % (projection.name,prune_sum,sprout_sum,(float(unit_total)/self.mask_total)*100))
def __call__(self, projection, **params):
time = math.ceil(topo.sim.time())
if self.disk_mask:
self.disk = ig.Disk(size=1.0,smoothing=0.0)
# Get CF and src sheet shapes
cf_x,cf_y = projection.dest.activity.shape
src_x,src_y = projection.src.activity.shape
# Initialize sparse triplet arrays
y_array = np.zeros((src_x*src_y*cf_y),dtype=np.int32)
x_array = np.zeros((src_x*src_y*cf_y),dtype=np.int32)
val_array = np.zeros((src_x*src_y*cf_y),dtype=sparse_type)
# Create new sparse matrix to accumulate into
sum_sparse = sparse.csarray_float(projection.src.activity.shape,projection.dest.activity.shape)
# Counters for logging
sprout_sum = 0; prune_sum = 0; unit_total = 0
self.mask_total = 0
if (time == 0):
if not hasattr(self,"initial_conns"):
self.initial_conns = {}
self.initial_conns[projection.name] = projection.n_conns()
elif (time % self.interval) == 0:
idx=0
for cidx,cf in enumerate(projection.flatcfs):
temp_weights = cf.weights
dense_unit_mask = (1.0 - (temp_weights>0.0))
dim1,dim2 = temp_weights.shape
sprout_count,prune_idx,nnz = self.calc_ratios(temp_weights)
self.prune(temp_weights,prune_idx)
nnz_pp = np.count_nonzero(temp_weights)
prune_sum += (nnz_pp-nnz)
if sprout_count:
self.sprout(temp_weights,dense_unit_mask,sprout_count)
nnz_ps = np.count_nonzero(temp_weights)
sprout_sum += nnz_ps - nnz_pp
unit_total += nnz_ps
# Populate sparse array chunk
temp_sparse = sparse.csarray_float(projection.src.activity.shape,projection.dest.activity.shape)
x1,x2,y1,y2 = cf.input_sheet_slice.tolist()
for cnx in range(dim1):
val_array[idx:idx+dim2] = temp_weights[cnx,:]
x_val = (x1+cnx) * src_y + y1
x_array[idx:idx+dim2] = range(x_val,x_val+dim2)
y_array[idx:idx+dim2] = cidx
idx += dim2
# Populate combined sparse array with sparse array chunk
if (cidx+1)%cf_y == 0:
nnz_idx = val_array.nonzero()
temp_sparse.setTriplets(x_array[nnz_idx],y_array[nnz_idx],val_array[nnz_idx])
sum_sparse += temp_sparse
x_array *= 0; y_array *= 0; val_array *= 0.0
idx=0
projection.weights = sum_sparse
del temp_sparse, sum_sparse
projection.weights.compress()
