def extend(self, keys, unitary=False):
"""Extends the vocabulary with additional keys.
Creates and adds the semantic pointers listed in keys to the
vocabulary.
Parameters
----------
keys : list
List of semantic pointer names to be added to the vocabulary.
unitary : bool or list, optional (Default: False)
If True, all generated pointers will be unitary. If a list of
strings, any pointer whose name is on the list will be forced to
be unitary when created.
"""
if is_iterable(unitary):
if is_iterable(self.unitary):
self.unitary.extend(unitary)
else:
self.unitary = list(unitary)
elif unitary:
if is_iterable(self.unitary):
self.unitary.extend(keys)
else:
self.unitary = list(keys)
for key in keys:
if key not in self.keys:
self[key]
def __init__(self, shape_in, filters, biases=None, stride=1, padding=0, activation='linear'): # noqa: C901
from nengo.utils.compat import is_iterable, is_integer
self.shape_in = tuple(shape_in)
self.filters = filters
self.stride = stride if is_iterable(stride) else [stride] * 2
self.padding = padding if is_iterable(padding) else [padding] * 2
self.activation=activation
nf = self.filters.shape[0]
nxi, nxj = self.shape_in[1:]
si, sj = self.filters.shape[-2:]
pi, pj = self.padding
sti, stj = self.stride
nyi = 1 + max(int(np.ceil((2*pi + nxi - si) / float(sti))), 0)
nyj = 1 + max(int(np.ceil((2*pj + nxj - sj) / float(stj))), 0)
self.shape_out = (nf, nyi, nyj)
self.biases = biases if biases is not None else None
if self.biases is not None:
if self.biases.size == 1:
self.biases.shape = (1, 1, 1)
elif self.biases.size == np.prod(self.shape_out):
self.biases.shape = self.shape_out
elif self.biases.size == self.shape_out[0]:
self.biases.shape = (self.shape_out[0], 1, 1)
elif self.biases.size == np.prod(self.shape_out[1:]):
self.biases.shape = (1,) + self.shape_out[1:]
super(Conv2d, self).__init__(
default_size_in=np.prod(self.shape_in),
default_size_out=np.prod(self.shape_out))
def add_input(self, name, input_vectors, input_scale=1.0):
# Handle different vocabulary types
if is_iterable(input_vectors):
input_vectors = np.matrix(input_vectors)
# Handle possible different input_scale values for each
# element in the associative memory
if not is_iterable(input_scale):
input_scale = np.matrix([input_scale] * input_vectors.shape[0])
else:
input_scale = np.matrix(input_scale)
if input_scale.shape[1] != input_vectors.shape[0]:
raise ValueError(
'Number of input_scale values do not match number of input '
'vectors. Got: %d, expected %d.' %
(input_scale.shape[1], input_vectors.shape[0]))
input = nengo.Node(size_in=input_vectors.shape[1], label=name)
if hasattr(self, name):
raise NameError('Name "%s" already exists as a node in the'
'associative memory.')
else:
setattr(self, name, input)
nengo.Connection(input, self.elem_input,
synapse=None,
transform=np.multiply(input_vectors, input_scale.T))
def __init__(self,
shape_in,
pool_size,
strides=None,
kind='avg',
mode='full'):
self.shape_in = shape_in
self.pool_size = (pool_size
if is_iterable(pool_size) else [pool_size] * 2)
self.strides = (strides if is_iterable(strides) else [strides] *
2 if strides is not None else self.pool_size)
self.kind = kind
self.mode = mode
if not all(st <= p for st, p in zip(self.strides, self.pool_size)):
raise ValueError("Strides %s must be <= pool_size %s" %
(self.strides, self.pool_size))
nc, nxi, nxj = self.shape_in
nyi_float = float(nxi - self.pool_size[0]) / self.strides[0]
nyj_float = float(nxj - self.pool_size[1]) / self.strides[1]
if self.mode == 'full':
nyi = 1 + int(np.ceil(nyi_float))
nyj = 1 + int(np.ceil(nyj_float))
elif self.mode == 'valid':
nyi = 1 + int(np.floor(nyi_float))
nyj = 1 + int(np.floor(nyj_float))
self.shape_out = (nc, nyi, nyj)
super(Pool2d, self).__init__(default_size_in=np.prod(self.shape_in),
default_size_out=np.prod(self.shape_out))
def add_input(self, name, input_vectors, input_scales=1.0):
# Handle different vocabulary types
if is_iterable(input_vectors):
input_vectors = np.matrix(input_vectors)
# Handle possible different input_scale values for each
# element in the associative memory
if not is_iterable(input_scales):
input_scales = np.matrix([input_scales] * input_vectors.shape[0])
else:
input_scales = np.matrix(input_scales)
if input_scales.shape[1] != input_vectors.shape[0]:
raise ValueError(
'Number of input_scale values do not match number of input '
'vectors. Got: %d, expected %d.' %
(input_scales.shape[1], input_vectors.shape[0]))
input = nengo.Node(size_in=input_vectors.shape[1], label=name)
if hasattr(self, name):
raise NameError('Name "%s" already exists as a node in the'
'associative memory.')
else:
setattr(self, name, input)
nengo.Connection(input, self.elem_input,
synapse=None,
transform=np.multiply(input_vectors, input_scales.T))
def __init__(self, shape_in, pool_size, strides=None,
kind='avg', mode='full'):
self.shape_in = shape_in
self.pool_size = (pool_size if is_iterable(pool_size) else
[pool_size] * 2)
self.strides = (strides if is_iterable(strides) else
[strides] * 2 if strides is not None else
self.pool_size)
self.kind = kind
self.mode = mode
if not all(st <= p for st, p in zip(self.strides, self.pool_size)):
raise ValueError("Strides %s must be <= pool_size %s" %
(self.strides, self.pool_size))
nc, nxi, nxj = self.shape_in
nyi_float = float(nxi - self.pool_size[0]) / self.strides[0]
nyj_float = float(nxj - self.pool_size[1]) / self.strides[1]
if self.mode == 'full':
nyi = 1 + int(np.ceil(nyi_float))
nyj = 1 + int(np.ceil(nyj_float))
elif self.mode == 'valid':
nyi = 1 + int(np.floor(nyi_float))
nyj = 1 + int(np.floor(nyj_float))
self.shape_out = (nc, nyi, nyj)
super(Pool2d, self).__init__(
default_size_in=np.prod(self.shape_in),
default_size_out=np.prod(self.shape_out))
def __init__(self,
shape_in,
filters,
biases=None,
strides=1,
padding=0,
border='ceil'): # noqa: C901
self.shape_in = shape_in
self.filters = filters
if self.filters.ndim not in [4, 6]:
raise ValueError(
"`filters` must have four or six dimensions "
"(filters, [height, width,] channels, f_height, f_width)")
if self.filters.shape[-3] != self.shape_in[0]:
raise ValueError(
"Filter channels (%d) and input channels (%d) must match" %
(self.filters.shape[-3], self.shape_in[0]))
if not all(s % 2 == 1 for s in self.filters.shape[-2:]):
raise ValueError("Filter shapes must be odd (got %r)" %
(self.filters.shape[-2:], ))
self.strides = strides if is_iterable(strides) else [strides] * 2
self.padding = padding if is_iterable(padding) else [padding] * 2
self.border = border
nf = self.filters.shape[0]
nxi, nxj = self.shape_in[1:]
si, sj = self.filters.shape[-2:]
pi, pj = self.padding
sti, stj = self.strides
rounder = np.ceil if self.border == 'ceil' else np.floor
nyi = 1 + max(int(rounder(float(2 * pi + nxi - si) / sti)), 0)
nyj = 1 + max(int(rounder(float(2 * pj + nxj - sj) / stj)), 0)
self.shape_out = (nf, nyi, nyj)
if self.filters.ndim == 6 and self.filters.shape[1:3] != (nyi, nyj):
raise ValueError("Number of local filters %r must match out shape "
"%r" % (self.filters.shape[1:3], (nyi, nyj)))
self.biases = biases if biases is not None else None
if self.biases is not None:
if self.biases.size == 1:
self.biases.shape = (1, 1, 1)
elif self.biases.size == np.prod(self.shape_out):
self.biases.shape = self.shape_out
elif self.biases.size == self.shape_out[0]:
self.biases.shape = (self.shape_out[0], 1, 1)
elif self.biases.size == np.prod(self.shape_out[1:]):
self.biases.shape = (1, ) + self.shape_out[1:]
else:
raise ValueError(
"Biases size (%d) does not match output shape %s" %
(self.biases.size, self.shape_out))
super(Conv2d, self).__init__(default_size_in=np.prod(self.shape_in),
default_size_out=np.prod(self.shape_out))
def simplify(self):
is_num = lambda x: isinstance(x, NumExp)
if isinstance(self.fn, str):
return self # cannot simplify
elif all(map(is_num, self.args)):
# simplify scalar function
return NumExp(self.fn(*[a.value for a in self.args]))
elif all(is_num(a) or is_iterable(a) and all(map(is_num, a))
for a in self.args):
# simplify vector function
return NumExp(self.fn(
[[aa.value for aa in a] if is_iterable(a) else a.value
for a in self.args]))
else:
return self # cannot simplify
def __init__(self, shape_in, filters, biases=None, strides=1, padding=0): # noqa: C901
self.shape_in = shape_in
self.filters = filters
if self.filters.ndim not in [4, 6]:
raise ValueError(
"`filters` must have four or six dimensions "
"(filters, [height, width,] channels, f_height, f_width)")
if self.filters.shape[-3] != self.shape_in[0]:
raise ValueError(
"Filter channels (%d) and input channels (%d) must match"
% (self.filters.shape[-3], self.shape_in[0]))
if not all(s % 2 == 1 for s in self.filters.shape[-2:]):
raise ValueError("Filter shapes must be odd (got %r)"
% (self.filters.shape[-2:],))
self.strides = strides if is_iterable(strides) else [strides] * 2
self.padding = padding if is_iterable(padding) else [padding] * 2
nf = self.filters.shape[0]
nxi, nxj = self.shape_in[1:]
si, sj = self.filters.shape[-2:]
pi, pj = self.padding
sti, stj = self.strides
nyi = 1 + max(int(np.ceil(float(2*pi + nxi - si) / sti)), 0)
nyj = 1 + max(int(np.ceil(float(2*pj + nxj - sj) / stj)), 0)
self.shape_out = (nf, nyi, nyj)
if self.filters.ndim == 6 and self.filters.shape[1:3] != (nyi, nyj):
raise ValueError("Number of local filters %r must match out shape "
"%r" % (self.filters.shape[1:3], (nyi, nyj)))
self.biases = biases if biases is not None else None
if self.biases is not None:
if self.biases.size == 1:
self.biases.shape = (1, 1, 1)
elif self.biases.size == np.prod(self.shape_out):
self.biases.shape = self.shape_out
elif self.biases.size == self.shape_out[0]:
self.biases.shape = (self.shape_out[0], 1, 1)
elif self.biases.size == np.prod(self.shape_out[1:]):
self.biases.shape = (1,) + self.shape_out[1:]
else:
raise ValueError(
"Biases size (%d) does not match output shape %s"
% (self.biases.size, self.shape_out))
super(Conv2d, self).__init__(
default_size_in=np.prod(self.shape_in),
default_size_out=np.prod(self.shape_out))
def validate(self, instance, rule):
if is_iterable(rule):
for r in (itervalues(rule) if isinstance(rule, dict) else rule):
self.validate_rule(instance, r)
elif rule is not None:
self.validate_rule(instance, rule)
super(LearningRuleTypeParam, self).validate(instance, rule)
def validate(self, instance, rule):
if is_iterable(rule):
for r in (itervalues(rule) if isinstance(rule, dict) else rule):
self.validate_rule(instance, r)
elif rule is not None:
self.validate_rule(instance, rule)
super(LearningRuleTypeParam, self).validate(instance, rule)
def __getitem__(self, item):
if isinstance(item, slice):
item = np.arange(len(self))[item]
if is_iterable(item):
rval = self.__class__.__new__(self.__class__)
rval.starts = [self.starts[i] for i in item]
rval.shape0s = [self.shape0s[i] for i in item]
rval.shape1s = [self.shape1s[i] for i in item]
rval.stride0s = [self.stride0s[i] for i in item]
rval.stride1s = [self.stride1s[i] for i in item]
rval.buf = self.buf
rval.names = [self.names[i] for i in item]
return rval
else:
if isinstance(item, np.ndarray):
item.shape = () # avoid numpy DeprecationWarning
itemsize = self.dtype.itemsize
shape = (self.shape0s[item], self.shape1s[item])
byteoffset = itemsize * self.starts[item]
bytestrides = (itemsize * self.stride0s[item],
itemsize * self.stride1s[item])
return np.ndarray(shape=shape,
dtype=self.dtype,
buffer=self.buf.data,
offset=byteoffset,
strides=bytestrides)
def coerce(self, instance, rule):
if is_iterable(rule):
for r in (itervalues(rule) if isinstance(rule, dict) else rule):
self.check_rule(instance, r)
elif rule is not None:
self.check_rule(instance, rule)
return super(LearningRuleTypeParam, self).coerce(instance, rule)
def __setitem__(self, item, new_value):
if isinstance(item, slice) or is_iterable(item):
raise NotImplementedError('TODO')
else:
m, n = self.shape0s[item], self.shape1s[item]
sm, sn = self.stride0s[item], self.stride1s[item]
if (sm, sn) in [(1, m), (n, 1)]:
# contiguous
clarray = self.getitem_device(item)
if isinstance(new_value, np.ndarray):
array = new_value.astype(self.dtype)
else:
array = np.zeros(clarray.shape, dtype=clarray.dtype)
array[...] = new_value
array.shape = clarray.shape # reshape to avoid warning
clarray.set(array)
else:
# discontiguous
# Copy a contiguous region off the device that surrounds the
# discontiguous, set the appropriate values, and copy back
s = self.starts[item]
array = to_host(self.queue, self.cl_buf.data, self.dtype,
s, (m, n), (sm, sn), is_blocking=True)
array[...] = new_value
buf = array.base if array.base is not None else array
bytestart = self.dtype.itemsize * s
cl.enqueue_copy(self.queue, self.cl_buf.data, buf,
device_offset=bytestart, is_blocking=True)
def validate(self, instance, rule):
if is_iterable(rule):
for lr in rule:
self.validate_rule(instance, lr)
elif rule is not None:
self.validate_rule(instance, rule)
super(LearningRuleParam, self).validate(instance, rule)
def __getitem__(self, item):
if isinstance(item, slice):
item = np.arange(len(self))[item]
if is_iterable(item):
rval = self.__class__.__new__(self.__class__)
rval.starts = [self.starts[i] for i in item]
rval.shape0s = [self.shape0s[i] for i in item]
rval.shape1s = [self.shape1s[i] for i in item]
rval.stride0s = [self.stride0s[i] for i in item]
rval.stride1s = [self.stride1s[i] for i in item]
rval.buf = self.buf
rval.names = [self.names[i] for i in item]
return rval
else:
if isinstance(item, np.ndarray):
item.shape = () # avoid numpy DeprecationWarning
itemsize = self.dtype.itemsize
shape = (self.shape0s[item], self.shape1s[item])
byteoffset = itemsize * self.starts[item]
bytestrides = (itemsize * self.stride0s[item],
itemsize * self.stride1s[item])
return np.ndarray(
shape=shape, dtype=self.dtype, buffer=self.buf.data,
offset=byteoffset, strides=bytestrides)
def __setitem__(self, item, new_value):
if isinstance(item, slice) or is_iterable(item):
raise NotImplementedError('TODO')
else:
m, n = self.shape0s[item], self.shape1s[item]
sm, sn = self.stride0s[item], self.stride1s[item]
if (sm, sn) in [(1, m), (n, 1)]:
# contiguous
clarray = self.getitem_device(item)
if isinstance(new_value, np.ndarray):
array = np.asarray(new_value, order='C', dtype=self.dtype)
else:
array = np.zeros(clarray.shape, dtype=clarray.dtype)
array[...] = new_value
array.shape = clarray.shape # reshape to avoid warning
assert equal_strides(
array.strides, clarray.strides, clarray.shape)
clarray.set(array)
else:
# discontiguous
# Copy a contiguous region off the device that surrounds the
# discontiguous, set the appropriate values, and copy back
s = self.starts[item]
array = to_host(self.queue, self.cl_buf.data, self.dtype,
s, (m, n), (sm, sn), is_blocking=True)
array[...] = new_value
buf = array.base if array.base is not None else array
bytestart = self.dtype.itemsize * s
cl.enqueue_copy(self.queue, self.cl_buf.data, buf,
device_offset=bytestart, is_blocking=True)
def __init__(
self, dimensions, strict=True, max_similarity=0.1,
pointer_gen=None, name=None, algebra=None):
if algebra is None:
algebra = HrrAlgebra()
self.algebra = algebra
if not is_integer(dimensions) or dimensions < 1:
raise ValidationError("dimensions must be a positive integer",
attr='dimensions', obj=self)
if pointer_gen is None:
pointer_gen = UnitLengthVectors(dimensions)
elif isinstance(pointer_gen, np.random.RandomState):
pointer_gen = UnitLengthVectors(dimensions, pointer_gen)
if not is_iterable(pointer_gen) or is_string(pointer_gen):
raise ValidationError(
"pointer_gen must be iterable or RandomState",
attr='pointer_gen', obj=self)
self.dimensions = dimensions
self.strict = strict
self.max_similarity = max_similarity
self._key2idx = {}
self._keys = []
self._vectors = np.zeros((0, dimensions), dtype=float)
self.pointer_gen = pointer_gen
self.name = name
def add_output(self, name, function, synapse=None, **conn_kwargs):
dims_per_ens = self.dimensions_per_ensemble
# get output size for each ensemble
sizes = np.zeros(self.n_ensembles, dtype=int)
if is_iterable(function) and all(callable(f) for f in function):
if len(list(function)) != self.n_ensembles:
raise ValueError("Must have one function per ensemble")
for i, func in enumerate(function):
sizes[i] = np.asarray(func(np.zeros(dims_per_ens))).size
elif callable(function):
sizes[:] = np.asarray(function(np.zeros(dims_per_ens))).size
function = [function] * self.n_ensembles
elif function is None:
sizes[:] = dims_per_ens
function = [None] * self.n_ensembles
else:
raise ValueError(
"'function' must be a callable, list of callables, or 'None'")
output = nengo.Node(output=None, size_in=sizes.sum(), label=name)
setattr(self, name, output)
indices = np.zeros(len(sizes) + 1, dtype=int)
indices[1:] = np.cumsum(sizes)
for i, e in enumerate(self.ea_ensembles):
nengo.Connection(e,
output[indices[i]:indices[i + 1]],
function=function[i],
synapse=synapse,
**conn_kwargs)
return output
def add_output_mapping(self, name, output_vectors):
"""Adds another output to the associative memory network.
Creates a transform with the given output vectors between the
associative memory element output and a named output node to enable the
selection of output vectors by the associative memory.
Parameters
----------
name: str
Name to use for the output node. This name will be used as
the name of the attribute for the associative memory network.
output_vectors: array_like
The list of vectors to be produced for each match.
"""
# --- Put arguments in canonical form
if is_iterable(output_vectors):
output_vectors = np.array(output_vectors, ndmin=2)
# --- Check preconditions
if hasattr(self, name):
raise ValidationError("Name '%s' already exists as a node in the "
"associative memory." % name, attr='name')
# --- Make the output node and connect it
output = nengo.Node(size_in=output_vectors.shape[1], label=name)
setattr(self, name, output)
if self.thresh_ens is not None:
c = nengo.Connection(self.thresh_ens.output, output,
synapse=None, transform=output_vectors.T)
else:
c = nengo.Connection(self.elem_output, output,
synapse=None, transform=output_vectors.T)
self.out_conns.append(c)
def add_output(self, name, function, synapse=None, **conn_kwargs):
dims_per_ens = self.dimensions_per_ensemble
# get output size for each ensemble
sizes = np.zeros(self.n_ensembles, dtype=int)
if is_iterable(function) and all(callable(f) for f in function):
if len(list(function)) != self.n_ensembles:
raise ValidationError(
"Must have one function per ensemble", attr='function')
for i, func in enumerate(function):
sizes[i] = np.asarray(func(np.zeros(dims_per_ens))).size
elif callable(function):
sizes[:] = np.asarray(function(np.zeros(dims_per_ens))).size
function = [function] * self.n_ensembles
elif function is None:
sizes[:] = dims_per_ens
function = [None] * self.n_ensembles
else:
raise ValidationError("'function' must be a callable, list of "
"callables, or None", attr='function')
output = nengo.Node(output=None, size_in=sizes.sum(), label=name)
setattr(self, name, output)
indices = np.zeros(len(sizes) + 1, dtype=int)
indices[1:] = np.cumsum(sizes)
for i, e in enumerate(self.ea_ensembles):
nengo.Connection(
e, output[indices[i]:indices[i+1]], function=function[i],
synapse=synapse, **conn_kwargs)
return output
def validate(self, instance, rule):
if is_iterable(rule):
for lr in rule:
self.validate_rule(instance, lr)
elif rule is not None:
self.validate_rule(instance, rule)
super(LearningRuleParam, self).validate(instance, rule)
def __init__(self, fn, in_dims=None, out_dim=None):
if in_dims is not None and not is_iterable(in_dims):
in_dims = [in_dims]
self.fn = fn
self.in_dims = in_dims
self.out_dim = out_dim
self._translator = None
def __init__(self, fn, in_dims=None, out_dim=None):
if in_dims is not None and not is_iterable(in_dims):
in_dims = [in_dims]
self.fn = fn
self.in_dims = in_dims
self.out_dim = out_dim
self._translator = None
def add_input_mapping(self, name, input_vectors, input_scales=1.0):
"""Adds a set of input vectors to the associative memory network.
Creates a transform with the given input vectors between the
a named input node and associative memory element input to enable the
inputs to be mapped onto ensembles of the Associative Memory.
Parameters
----------
name: string
Name to use for the input node. This name will be used as the name
of the attribute for the associative memory network.
input_vectors: array_like
The list of vectors to be compared against.
input_scales: array_list, optional
Scaling factor to apply on each of the input vectors. Note that it
is possible to scale each vector independently.
"""
# --- Put arguments in canonical form
n_vectors, d_vectors = input_vectors.shape
if is_iterable(input_vectors):
input_vectors = np.array(input_vectors, ndmin=2)
if not is_iterable(input_scales):
input_scales = input_scales * np.ones((1, n_vectors))
else:
input_scales = np.array(input_scales, ndmin=2)
# --- Check some preconditions
if input_scales.shape[1] != n_vectors:
raise ValidationError("Number of input_scale values (%d) does not "
"match number of input vectors (%d)." %
(input_scales.shape[1], n_vectors),
attr='input_scales')
if hasattr(self, name):
raise ValidationError("Name '%s' already exists as a node in the "
"associative memory." % name,
attr='name')
# --- Finally, make the input node and connect it
in_node = nengo.Node(size_in=d_vectors, label=name)
setattr(self, name, in_node)
nengo.Connection(in_node,
self.elem_input,
synapse=None,
transform=input_vectors * input_scales.T)
def __init__(self,
n_neurons,
n_ensembles,
ens_dimensions=1,
neuron_nodes=False,
label=None,
seed=None,
add_to_container=None,
**ens_kwargs):
if "dimensions" in ens_kwargs:
raise ValidationError(
"'dimensions' is not a valid argument to EnsembleArray. "
"To set the number of ensembles, use 'n_ensembles'. To set "
"the number of dimensions per ensemble, use 'ens_dimensions'.",
attr='dimensions',
obj=self)
super(EnsembleArray, self).__init__(label, seed, add_to_container)
for param in ens_kwargs:
if is_iterable(ens_kwargs[param]):
ens_kwargs[param] = nengo.dists.Samples(ens_kwargs[param])
self.config[nengo.Ensemble].update(ens_kwargs)
label_prefix = "" if label is None else label + "_"
self.n_neurons = n_neurons
self.n_ensembles = n_ensembles
self.dimensions_per_ensemble = ens_dimensions
# These may be set in add_neuron_input and add_neuron_output
self.neuron_input, self.neuron_output = None, None
self.ea_ensembles = []
with self:
self.input = nengo.Node(size_in=self.dimensions, label="input")
for i in range(n_ensembles):
e = nengo.Ensemble(n_neurons,
self.dimensions_per_ensemble,
label="%s%d" % (label_prefix, i))
nengo.Connection(self.input[i * ens_dimensions:(i + 1) *
ens_dimensions],
e,
synapse=None)
self.ea_ensembles.append(e)
if neuron_nodes:
self.add_neuron_input()
self.add_neuron_output()
warnings.warn(
"'neuron_nodes' argument will be removed in Nengo 2.2. Use "
"'add_neuron_input' and 'add_neuron_output' methods instead.",
DeprecationWarning)
self.add_output('output', function=None)
def add_input_mapping(self, name, input_vectors, input_scales=1.0):
"""Adds a set of input vectors to the associative memory network.
Creates a transform with the given input vectors between the
a named input node and associative memory element input to enable the
inputs to be mapped onto ensembles of the Associative Memory.
Parameters
----------
name: string
Name to use for the input node. This name will be used as the name
of the attribute for the associative memory network.
input_vectors: array_like
The list of vectors to be compared against.
input_scales: array_list, optional
Scaling factor to apply on each of the input vectors. Note that it
is possible to scale each vector independently.
"""
# --- Put arguments in canonical form
n_vectors, d_vectors = input_vectors.shape
if is_iterable(input_vectors):
input_vectors = np.array(input_vectors, ndmin=2)
if not is_iterable(input_scales):
input_scales = input_scales * np.ones((1, n_vectors))
else:
input_scales = np.array(input_scales, ndmin=2)
# --- Check some preconditions
if input_scales.shape[1] != n_vectors:
raise ValidationError("Number of input_scale values (%d) does not "
"match number of input vectors (%d)."
% (input_scales.shape[1], n_vectors),
attr='input_scales')
if hasattr(self, name):
raise ValidationError("Name '%s' already exists as a node in the "
"associative memory." % name, attr='name')
# --- Finally, make the input node and connect it
in_node = nengo.Node(size_in=d_vectors, label=name)
setattr(self, name, in_node)
nengo.Connection(in_node, self.elem_input,
synapse=None,
transform=input_vectors * input_scales.T)
def make_step(self, shape_in, shape_out, dt, rng):
size_out = shape_out[0] if is_iterable(shape_out) else shape_out
if self.base_output is None:
f = self.passthrough
elif isinstance(self.base_output, Process):
f = self.base_output.make_step(shape_in, shape_out, dt, rng)
else:
f = self.base_output
return self.Step(size_out, f, self.fast_client)
def function(self, _function):
if _function is not None:
self._check_pre_ensemble('function')
x = (self.eval_points[0] if is_iterable(self.eval_points) else
np.zeros(self._pre.dimensions))
size = np.asarray(_function(x)).size
else:
size = 0
self._function = (_function, size)
self._check_shapes()
def make_step(self, shape_in, shape_out, dt, rng):
size_out = shape_out[0] if is_iterable(shape_out) else shape_out
if self.base_output is None:
f = self.passthrough
elif isinstance(self.base_output, Process):
f = self.base_output.make_step(shape_in, shape_out, dt, rng)
else:
f = self.base_output
return self.Step(size_out, f,
to_client=self.to_client,
from_client=self.from_client)
def _broadcast_args(self, func, args):
"""Apply 'func' element-wise to lists of args"""
as_list = lambda x: list(x) if is_iterable(x) else [x]
args = list(map(as_list, args))
arg_lens = list(map(len, args))
max_len = max(arg_lens)
assert all(n in [0, 1, max_len] for n in arg_lens), (
"Could not broadcast arguments with lengths %s" % arg_lens)
result = [func(*[a[i] if len(a) > 1 else a[0] for a in args])
for i in range(max_len)]
result = [r.simplify() for r in result]
return result[0] if len(result) == 1 else result
def lsuv(X, ws, fs, **kwargs):
"""Layer-sequential unit-variance initialization [1]_
References
----------
.. [1] Mishkin, D., & Matas, J. (2016). All you need is a good init.
In ICLR 2016 (pp. 1-13).
"""
fs = ([fs] * (len(ws) - 1) + [None]) if not is_iterable(fs) else fs
assert len(ws) >= 2
assert len(fs) == len(ws)
for i, (w, f) in enumerate(zip(ws, fs)):
X = lsuv_layer(X, w, f, layer_i=i, **kwargs)
def visit_Return(self, expr):
value = self.visit(expr.value)
if is_iterable(value):
self._check_vector_length(len(value))
if not all(isinstance(v, Expression) for v in value):
raise ValueError(
"Can only return a list of mathematical expressions")
return ["%s[%d] = %s;" % (OUTPUT_NAME, i, v.to_ocl())
for i, v in enumerate(value)] + ["return;"]
elif isinstance(value, Expression):
return ["%s[0] = %s;" % (OUTPUT_NAME, value.to_ocl()), "return;"]
else:
raise ValueError("Can only return mathematical expressions, "
"or lists of expressions")
def learning_rule(self):
if self.learning_rule_type is not None and self._learning_rule is None:
types = self.learning_rule_type
if isinstance(types, dict):
self._learning_rule = types.__class__() # dict of same type
for k, v in iteritems(types):
self._learning_rule[k] = LearningRule(self, v)
elif is_iterable(types):
self._learning_rule = [LearningRule(self, v) for v in types]
elif isinstance(types, LearningRuleType):
self._learning_rule = LearningRule(self, types)
else:
raise ValueError("Invalid type for `learning_rule_type`: %s" %
(types.__class__.__name__))
return self._learning_rule
def __getitem__(self, item):
"""
Getting one item returns a numpy array (on the host).
Getting multiple items returns a view into the device.
"""
if is_iterable(item):
return self.getitem_device(item)
else:
buf = to_host(
self.queue, self.cl_buf.data, self.dtype, self.starts[item],
(self.shape0s[item], self.shape1s[item]),
(self.stride0s[item], self.stride1s[item]),
)
buf.setflags(write=False)
return buf
def learning_rule(self):
if self.learning_rule_type is not None and self._learning_rule is None:
types = self.learning_rule_type
if isinstance(types, dict):
self._learning_rule = types.__class__() # dict of same type
for k, v in iteritems(types):
self._learning_rule[k] = LearningRule(self, v)
elif is_iterable(types):
self._learning_rule = [LearningRule(self, v) for v in types]
elif isinstance(types, LearningRuleType):
self._learning_rule = LearningRule(self, types)
else:
raise ValueError("Invalid type for `learning_rule_type`: %s"
% (types.__class__.__name__))
return self._learning_rule
def __getitem__(self, item):
"""
Getting one item returns a numpy array (on the host).
Getting multiple items returns a view into the device.
"""
if is_iterable(item):
return self.getitem_device(item)
else:
buf = to_host(
self.queue, self.cl_buf.data, self.dtype, self.starts[item],
(self.shape0s[item], self.shape1s[item]),
(self.stride0s[item], self.stride1s[item]),
)
buf.setflags(write=False)
return buf
def similarity(data, vocab, normalize=False):
"""Return the similarity between simulation data and Semantic Pointers.
Computes the dot products between all Semantic Pointers in the Vocabulary
and the simulation data for each timestep. If ``normalize=True``,
normalizes all vectors to compute the cosine similarity.
Parameters
----------
data: (D,) or (T, D) array_like
The *D*-dimensional data for *T* timesteps used for comparison.
vocab: Vocabulary or array_like
Vocabulary (or list of vectors) used to calculate the similarity
values.
normalize : bool, optional
Whether to normalize all vectors, to compute the cosine similarity.
"""
if isinstance(data, SemanticPointer):
data = data.v
if isinstance(vocab, Vocabulary):
vectors = vocab.vectors
elif is_iterable(vocab):
if isinstance(next(iter(vocab)), SemanticPointer):
vocab = [p.v for p in vocab]
vectors = np.array(vocab, copy=False, ndmin=2)
else:
raise ValidationError("%r object is not a valid vocabulary" %
(type(vocab).__name__),
attr='vocab')
dots = np.dot(vectors, data.T)
if normalize:
# Zero-norm vectors should return zero, so avoid divide-by-zero error
eps = np.nextafter(0, 1) # smallest float above zero
dnorm = np.maximum(npext.norm(data.T, axis=0, keepdims=True), eps)
vnorm = np.maximum(npext.norm(vectors, axis=1, keepdims=True), eps)
if len(dots.shape) == 1:
vnorm = np.squeeze(vnorm)
dots /= dnorm
dots /= vnorm
return dots.T
def getitem_device(self, item):
if isinstance(item, slice):
item = np.arange(len(self))[item]
if is_iterable(item):
return CLRaggedArray.from_buffer(
self.queue, self.cl_buf, self.starts[item],
self.shape0s[item], self.shape1s[item],
self.stride0s[item], self.stride1s[item],
names=[self.names[i] for i in item])
else:
s = self.dtype.itemsize
return Array(
self.queue,
(self.shape0s[item], self.shape1s[item]), self.dtype,
strides=(self.stride0s[item] * s, self.stride1s[item] * s),
data=self.cl_buf.data, offset=self.starts[item] * s)
def __getitem__(self, key):
"""Return the semantic pointer with the requested name.
If one does not exist, automatically create one. The key must be
a valid semantic pointer name, which is any Python identifier starting
with a capital letter.
"""
if not key[0].isupper():
raise KeyError("Semantic pointers must begin with a capital")
value = self.pointers.get(key, None)
if value is None:
if is_iterable(self.unitary):
unitary = key in self.unitary
else:
unitary = self.unitary
value = self.create_pointer(unitary=unitary)
self.add(key, value)
return value
def learning_rule(self):
"""(LearningRule or iterable) Connectable learning rule object(s)."""
if self.learning_rule_type is not None and self._learning_rule is None:
types = self.learning_rule_type
if isinstance(types, dict):
self._learning_rule = types.__class__() # dict of same type
for k, v in iteritems(types):
self._learning_rule[k] = LearningRule(self, v)
elif is_iterable(types):
self._learning_rule = [LearningRule(self, v) for v in types]
elif isinstance(types, LearningRuleType):
self._learning_rule = LearningRule(self, types)
else:
raise ValidationError(
"Invalid type %r" % types.__class__.__name__,
attr='learning_rule_type', obj=self)
return self._learning_rule
def __getitem__(self, key):
"""Return the semantic pointer with the requested name.
If one does not exist, automatically create one. The key must be
a valid semantic pointer name, which is any Python identifier starting
with a capital letter.
"""
if not key[0].isupper():
raise KeyError('Semantic pointers must begin with a capital')
value = self.pointers.get(key, None)
if value is None:
if is_iterable(self.unitary):
unitary = key in self.unitary
else:
unitary = self.unitary
value = self.create_pointer(unitary=unitary)
self.add(key, value)
return value
def learning_rule(self):
"""(LearningRule or iterable) Connectable learning rule object(s)."""
if self.learning_rule_type is not None and self._learning_rule is None:
types = self.learning_rule_type
if isinstance(types, dict):
self._learning_rule = types.__class__() # dict of same type
for k, v in iteritems(types):
self._learning_rule[k] = LearningRule(self, v)
elif is_iterable(types):
self._learning_rule = [LearningRule(self, v) for v in types]
elif isinstance(types, LearningRuleType):
self._learning_rule = LearningRule(self, types)
else:
raise ValidationError("Invalid type %r" %
types.__class__.__name__,
attr='learning_rule_type',
obj=self)
return self._learning_rule
def add_output(self, name, output_vectors):
# Handle different vocabulary types
if is_iterable(output_vectors):
output_vectors = np.matrix(output_vectors)
output = nengo.Node(size_in=output_vectors.shape[1], label=name)
if hasattr(self, name):
raise NameError('Name "%s" already exists as a node in the'
'associative memory.')
else:
setattr(self, name, output)
if self.threshold_output:
nengo.Connection(self.thresh_ens.output, output, synapse=None,
transform=output_vectors.T)
else:
nengo.Connection(self.elem_output, output, synapse=None,
transform=output_vectors.T)
def add_output(self, name, output_vectors):
# Handle different vocabulary types
if is_iterable(output_vectors):
output_vectors = np.matrix(output_vectors)
output = nengo.Node(size_in=output_vectors.shape[1], label=name)
if hasattr(self, name):
raise NameError('Name "%s" already exists as a node in the'
'associative memory.')
else:
setattr(self, name, output)
if self.threshold_output:
nengo.Connection(self.thresh_ens.output, output, synapse=None,
transform=output_vectors.T)
else:
nengo.Connection(self.elem_output, output, synapse=None,
transform=output_vectors.T)
def __init__(self, systems, dt=None, elementwise=False, method='zoh'):
if not is_iterable(systems) or isinstance(systems, LinearSystem):
systems = [systems]
self.systems = systems
self.dt = dt
self.elementwise = elementwise
self.A = []
self.B = []
self.C = []
self.D = []
for sys in systems:
sys = LinearSystem(sys)
if dt is not None:
sys = cont2discrete(sys, dt, method=method)
elif sys.analog:
raise ValueError(
"system (%s) must be digital if not given dt" % sys)
A, B, C, D = sys.ss
self.A.append(A)
self.B.append(B)
self.C.append(C)
self.D.append(D)
# TODO: If all of the synapses are single order, than A is diagonal
# and so np.dot(self.A, self._x) is trivial. But perhaps
# block_diag is already optimized for this.
# Note: ideally we could put this into CCF to reduce the A mapping
# to a single dot product and a shift operation. But in general
# since this is MIMO it is not controllable from a single input.
# Instead we might want to consider balanced reduction to
# improve efficiency.
self.A = block_diag(*self.A)
self.B = block_diag(*self.B) if elementwise else np.vstack(self.B)
self.C = block_diag(*self.C)
self.D = block_diag(*self.D) if elementwise else np.vstack(self.D)
# TODO: shape validation
self._x = np.zeros(len(self.A))[:, None]
def similarity(data, vocab, normalize=False):
"""Return the similarity between some data and the vocabulary.
Computes the dot products between all data vectors and each
vocabulary vector. If `normalize=True`, normalizes all vectors
to compute the cosine similarity.
Parameters
----------
data: array_like
The data used for comparison.
vocab: spa.Vocabulary, array_like
Vocabulary (or list of vectors) to use to calculate
the similarity values
normalize : boolean (optional)
Whether to normalize all vectors, to compute the cosine similarity.
"""
from nengo.spa.vocab import Vocabulary
if isinstance(vocab, Vocabulary):
vectors = vocab.vectors
elif is_iterable(vocab):
vectors = np.array(vocab, copy=False, ndmin=2)
else:
raise ValidationError("%r object is not a valid vocabulary" %
(vocab.__class__.__name__),
attr='vocab')
data = np.array(data, copy=False, ndmin=2)
dots = np.dot(data, vectors.T)
if normalize:
# Zero-norm vectors should return zero, so avoid divide-by-zero error
eps = np.nextafter(0, 1) # smallest float above zero
dnorm = np.maximum(npext.norm(data, axis=1, keepdims=True), eps)
vnorm = np.maximum(npext.norm(vectors, axis=1, keepdims=True), eps)
dots /= dnorm
dots /= vnorm.T
return dots
def __init__(self, systems, dt=None, elementwise=False, method='zoh'):
if not is_iterable(systems):
systems = [systems]
self.systems = systems
self.dt = dt
self.elementwise = elementwise
self.A = []
self.B = []
self.C = []
self.D = []
for sys in systems:
sys = LinearSystem(sys)
if dt is not None:
sys = cont2discrete(sys, dt, method=method)
elif sys.analog:
raise ValueError(
"system (%s) must be digital if not given dt" % sys)
A, B, C, D = sys.ss
self.A.append(A)
self.B.append(B)
self.C.append(C)
self.D.append(D)
# TODO: If all of the synapses are single order, than A is diagonal
# and so np.dot(self.A, self._x) is trivial. But perhaps
# block_diag is already optimized for this.
# Note: ideally we could put this into CCF to reduce the A mapping
# to a single dot product and a shift operation. But in general
# since this is MIMO it is not controllable from a single input.
# Instead we might want to consider balanced reduction to
# improve efficiency.
self.A = block_diag(*self.A)
self.B = block_diag(*self.B) if elementwise else np.vstack(self.B)
self.C = block_diag(*self.C)
self.D = block_diag(*self.D) if elementwise else np.vstack(self.D)
# TODO: shape validation
self._x = np.zeros(len(self.A))[:, None]
def add_output_mapping(self, name, output_vectors):
"""Adds another output to the associative memory network.
Creates a transform with the given output vectors between the
associative memory element output and a named output node to enable the
selection of output vectors by the associative memory.
Parameters
----------
name: string
Name to use for the output node. This name will be used as
the name of the attribute for the associative memory network.
output_vectors: array_like
The list of vectors to be produced for each match.
"""
# --- Put arguments in canonical form
if is_iterable(output_vectors):
output_vectors = np.array(output_vectors, ndmin=2)
# --- Check preconditions
if hasattr(self, name):
raise ValidationError("Name '%s' already exists as a node in the "
"associative memory." % name,
attr='name')
# --- Make the output node and connect it
output = nengo.Node(size_in=output_vectors.shape[1], label=name)
setattr(self, name, output)
if self.thresh_ens is not None:
c = nengo.Connection(self.thresh_ens.output,
output,
synapse=None,
transform=output_vectors.T)
else:
c = nengo.Connection(self.elem_output,
output,
synapse=None,
transform=output_vectors.T)
self.out_conns.append(c)
def similarity(data, vocab, normalize=False):
"""Return the similarity between some data and the vocabulary.
Computes the dot products between all data vectors and each
vocabulary vector. If `normalize=True`, normalizes all vectors
to compute the cosine similarity.
Parameters
----------
data: array_like
The data used for comparison.
vocab: spa.Vocabulary, array_like
Vocabulary (or list of vectors) to use to calculate
the similarity values
normalize : boolean (optional)
Whether to normalize all vectors, to compute the cosine similarity.
"""
from nengo.spa.vocab import Vocabulary
if isinstance(vocab, Vocabulary):
vectors = vocab.vectors
elif is_iterable(vocab):
vectors = np.array(vocab, copy=False, ndmin=2)
else:
raise ValidationError("%r object is not a valid vocabulary"
% (vocab.__class__.__name__), attr='vocab')
data = np.array(data, copy=False, ndmin=2)
dots = np.dot(data, vectors.T)
if normalize:
# Zero-norm vectors should return zero, so avoid divide-by-zero error
eps = np.nextafter(0, 1) # smallest float above zero
dnorm = np.maximum(npext.norm(data, axis=1, keepdims=True), eps)
vnorm = np.maximum(npext.norm(vectors, axis=1, keepdims=True), eps)
dots /= dnorm
dots /= vnorm.T
return dots
def getitem_device(self, item):
if isinstance(item, slice):
item = np.arange(len(self))[item]
if is_iterable(item):
return CLRaggedArray.from_buffer(
self.queue,
self.cl_buf,
self.starts[item],
self.shape0s[item],
self.shape1s[item],
self.stride0s[item],
self.stride1s[item],
names=[self.names[i] for i in item])
else:
s = self.dtype.itemsize
return Array(self.queue, (self.shape0s[item], self.shape1s[item]),
self.dtype,
strides=(self.stride0s[item] * s,
self.stride1s[item] * s),
data=self.cl_buf.data,
offset=self.starts[item] * s)
def getitem_device(self, item):
if isinstance(item, slice):
item = np.arange(len(self))[item]
if is_iterable(item):
rval = self.__class__.__new__(self.__class__)
rval.queue = self.queue
rval.starts = self.starts[item]
rval.shape0s = self.shape0s[item]
rval.shape1s = self.shape1s[item]
rval.stride0s = self.stride0s[item]
rval.stride1s = self.stride1s[item]
rval.cl_buf = self.cl_buf
rval.names = [self.names[i] for i in item]
return rval
else:
s = self.dtype.itemsize
return Array(
self.queue,
(self.shape0s[item], self.shape1s[item]), self.dtype,
strides=(self.stride0s[item] * s, self.stride1s[item] * s),
data=self.cl_buf.data, offset=self.starts[item] * s)
def similarity(data, probe, vocab=None):
"""Return the similarity between the probed data and the vocabulary.
Parameters
----------
data: ProbeDict
Collection of simulation data returned by sim.run() function call.
probe: Probe
Probe with desired data.
vocab: spa.Vocabulary, list, np.ndarray, np.matrix, optional
Optional vocabulary (or list of vectors) to use to calculate
the similarity values
"""
if vocab is None:
probe_vectors = probe.target.vocab.vectors.T
elif isinstance(vocab, Vocabulary):
probe_vectors = vocab.vectors.T
elif is_iterable(vocab):
probe_vectors = np.matrix(vocab).T
else:
probe_vectors = vocab.T
return np.dot(data[probe], probe_vectors)
def build_network(model, network):
"""Takes a Network object and returns a Model.
This determines the signals and operators necessary to simulate that model.
Builder does this by mapping each high-level object to its associated
signals and operators one-by-one, in the following order:
1) Ensembles, Nodes, Neurons
2) Subnetworks (recursively)
3) Connections
4) Learning Rules
5) Probes
"""
def get_seed(obj, rng):
# Generate a seed no matter what, so that setting a seed or not on
# one object doesn't affect the seeds of other objects.
seed = rng.randint(npext.maxint)
return (seed if not hasattr(obj, 'seed') or obj.seed is None
else obj.seed)
if model.toplevel is None:
model.toplevel = network
model.sig['common'][0] = Signal(
npext.array(0.0, readonly=True), name='Common: Zero')
model.sig['common'][1] = Signal(
npext.array(1.0, readonly=True), name='Common: One')
model.seeds[network] = get_seed(network, np.random)
# Set config
old_config = model.config
model.config = network.config
# assign seeds to children
rng = np.random.RandomState(model.seeds[network])
sorted_types = sorted(network.objects, key=lambda t: t.__name__)
for obj_type in sorted_types:
for obj in network.objects[obj_type]:
model.seeds[obj] = get_seed(obj, rng)
logger.debug("Network step 1: Building ensembles and nodes")
for obj in network.ensembles + network.nodes:
model.build(obj)
logger.debug("Network step 2: Building subnetworks")
for subnetwork in network.networks:
model.build(subnetwork)
logger.debug("Network step 3: Building connections")
for conn in network.connections:
model.build(conn)
logger.debug("Network step 4: Building learning rules")
for conn in network.connections:
rule = conn.learning_rule
if is_iterable(rule):
for r in (itervalues(rule) if isinstance(rule, dict) else rule):
model.build(r)
elif rule is not None:
model.build(rule)
logger.debug("Network step 5: Building probes")
for probe in network.probes:
model.build(probe)
# Unset config
model.config = old_config
model.params[network] = None
def function_args(self, conn, function):
x = (conn.eval_points[0] if is_iterable(conn.eval_points)
else np.zeros(conn.size_in))
return (x,)
