def predict_loop(self,
inputData,
outputData1,
outputData2,
continuation=True,
initialData=None,
update_processor=lambda x: x,
verbose=0):
inputData = B.array(inputData)
#let some input run through the ESN to initialize its states from a new starting value
if (not continuation):
self._esn1._x = B.zeros(self._esn1._x.shape)
self._esn2._x = B.zeros(self._esn2._x.shape)
total_length = inputData.shape[0]
aggregated_y1 = B.empty((total_length, self._n_output1))
aggregated_y2 = B.empty((total_length, self._n_output2))
# X1, y1 = self._esn1.propagate(inputData=inputData, outputData=None, transientTime=self._transientTime, verbose=verbose-1)
# Y1 = B.dot(self._esn1._WOut, X1)
# Y1 = update_processor(self.out_activation(Y1)).T
y2 = self.out_activation(B.dot(self._esn2._WOut, self._esn2._X).T)
for i in range(total_length):
inputDataWithFeedback = B.zeros((self.n_input + self._n_output2))
inputDataWithFeedback[:self.n_input] = inputData[i, :]
inputDataWithFeedback[self.n_input:] = y2
#update models
X1, _ = self._esn1.propagateOneStep(
inputData=inputDataWithFeedback,
outputData=None,
step=i,
transientTime=self._transientTime,
verbose=verbose - 1,
learn=False)
y1 = B.dot(self._esn1._WOut, X1)
aggregated_y1[i, :] = update_processor(self.out_activation(y1)).T
#output from 1st layer and correct output
#input2 = outputData1[i] - y1.reshape(self._n_output1)
# input2 = B.vstack((y1.reshape(self._n_output1), outputData1[i]))
# x2, y2 = self._esn2.propagateOneStep(inputData=input2, outputData=None, step=i, transientTime=self._transientTime, verbose=verbose-1, learn=False)
# Y_target2[i,:] = y2.reshape(self._n_output2)
training_error1 = B.sqrt(B.mean((aggregated_y1 - outputData1)**2))
print("training errors")
print(training_error1)
return aggregated_y1, aggregated_y2
def _fitProcess(self, data):
try:
inData, outData, indices, state = data
transientTime = self.sharedNamespace.transientTime
partialLength = self.sharedNamespace.partialLength
totalLength = self.sharedNamespace.totalLength
timeseriesCount = self.sharedNamespace.timeseriesCount
workerID = self.parallelWorkerIDs.get()
self._x[workerID] = state
# propagate
X = B.empty((1 + self.n_input + self.n_reservoir, totalLength))
for i in range(timeseriesCount):
X[:, i * partialLength:(i + 1) * partialLength] = self.propagate(inData[i], transientTime=transientTime,
x=self._x[workerID], verbose=0)
# define the target values
Y_target = B.empty((1, totalLength))
for i in range(timeseriesCount):
Y_target[:, i * partialLength:(i + 1) * partialLength] = self.out_inverse_activation(outData[i]).T[:,
transientTime:]
# now fit
WOut = None
if self._solver == "pinv":
WOut = B.dot(Y_target, B.pinv(X))
elif self._solver == "lsqr":
X_T = X.T
WOut = B.dot(B.dot(Y_target, X_T), B.inv(
B.dot(X, X_T) + self._regressionParameters[0] * B.identity(1 + self.n_input + self.n_reservoir)))
# calculate the training prediction now
# trainingPrediction = self.out_activation(B.dot(WOut, X).T)
# store the state and the output matrix of the worker
SpatioTemporalESN._fitProcess.fitQueue.put(
([x - self._filterWidth for x in indices], self._x[workerID].copy(), WOut.copy()))
self.parallelWorkerIDs.put(workerID)
except Exception as ex:
print(ex)
import traceback
traceback.print_exc()
SpatioTemporalESN._fitProcess.fitQueue.put(([x - self._filterWidth for x in indices], None, None))
self.parallelWorkerIDs.put(workerID)
def calculateTransientTime(self, inputs, outputs, epsilon, proximityLength = None):
# inputs: input of reserovoir
# outputs: output of reservoir
# epsilon: given constant
# proximity length: number of steps for which all states have to be epsilon close to declare convergance
# initializes two initial states as far as possible from each other in [-1,1] regime and tests when they converge-> this is transient time
length = inputs.shape[0] if inputs is not None else outputs.shape[0]
if proximityLength is None:
proximityLength = int(length * 0.1)
if proximityLength < 3:
proximityLength = 3
initial_x = B.empty((2, self.n_reservoir, 1))
initial_x[0] = - B.ones((self.n_reservoir, 1))
initial_x[1] = B.ones((self.n_reservoir, 1))
countedConsecutiveSteps = 0
length = inputs.shape[0] if inputs is not None else outputs.shape[0]
for t in range(length):
if B.max(B.ptp(initial_x, axis=0)) < epsilon:
if countedConsecutiveSteps >= proximityLength:
return t - proximityLength
else:
countedConsecutiveSteps += 1
else:
countedConsecutiveSteps = 0
u = inputs[t].reshape(-1, 1) if inputs is not None else None
o = outputs[t].reshape(-1, 1) if outputs is not None else None
for i in range(initial_x.shape[0]):
self.update(u, o, initial_x[i])
#transient time could not be determined
raise ValueError("Transient time could not be determined - maybe the proximityLength is too big.")
def getEquilibriumState(inputs, outputs, epsilon = 1e-3):
# inputs: input of reserovoir
# outputs: output of reservoir
# epsilon: given constant
# returns the equilibrium state when esn is fed with the first state of input
x = B.empty((2, self.n_reservoir, 1))
while not B.max(B.ptp(x, axis=0)) < epsilon:
x[0] = x[1]
u = inputs[0].reshape(-1, 1) if inputs is not None else None
o = outputs[0].reshape(-1, 1) if outputs is not None else None
self.update(u, o, x[1])
return x[1]
def predict(self,
inputData,
outputData1,
outputData2,
continuation=True,
initialData=None,
update_processor=lambda x: x,
verbose=0):
inputData = B.array(inputData)
#let some input run through the ESN to initialize its states from a new starting value
if (not continuation):
self._esn1._x = B.zeros(self._esn1._x.shape)
self._esn2._x = B.zeros(self._esn2._x.shape)
total_length = inputData.shape[0]
aggregated_y1 = B.empty((total_length, self._n_output1))
aggregated_y2 = B.empty((total_length, self._n_output2))
#put pixels and switch data together
inputDataWithFeedback = B.zeros(
(total_length, self.n_input + self._n_output2))
inputDataWithFeedback[:, :self.n_input] = inputData
inputDataWithFeedback[:, self.n_input:] = outputData2
X1, _ = self._esn1.propagate(inputData=inputDataWithFeedback,
outputData=None,
transientTime=self._transientTime,
verbose=verbose - 1)
aggregated_y1 = B.dot(self._esn1._WOut, X1)
aggregated_y1 = update_processor(self.out_activation(aggregated_y1)).T
training_error1 = B.sqrt(B.mean((aggregated_y1 - outputData1)**2))
print("predict errors")
print(training_error1)
return aggregated_y1, aggregated_y2
def predict(
self, inputData, update_processor=lambda x: x, transientTime=0, verbose=0
):
if len(inputData.shape) == 1:
inputData = inputData[None, :]
predictionLength = inputData.shape[1]
Y = B.empty((inputData.shape[0], self.n_output))
if verbose > 0:
bar = progressbar.ProgressBar(
max_value=inputData.shape[0], redirect_stdout=True, poll_interval=0.0001
)
bar.update(0)
for n in range(inputData.shape[0]):
# reset the state
self._x = B.zeros(self._x.shape)
X = self.propagate(inputData[n], transientTime)
# calculate the prediction using the trained model
if self._solver in [
"sklearn_auto",
"sklearn_lsqr",
"sklearn_sag",
"sklearn_svd",
"sklearn_svr",
]:
y = self._ridgeSolver.predict(X.T).reshape((self.n_output, -1))
else:
y = B.dot(self._W_out, X)
Y[n] = np.mean(y, 1)
if verbose > 0:
bar.update(n)
if verbose > 0:
bar.finish()
# return the result
result = B.zeros(Y.shape)
for i, n in enumerate(B.argmax(Y, 1)):
result[i, n] = 1.0
return result
def update(self, inputData, outputData=None, x=None):
if x is None:
x = self._x
if self._WFeedback is None:
# reshape the data
u = inputData.reshape(self.n_input, 1)
# update the states
transmission = self.calculateLinearNetworkTransmissions(u, x)
x *= 1.0 - self._leakingRate
x += self._leakingRate * self._activation(
transmission + (np.random.rand() - 0.5) * self._noiseLevel)
return u
else:
# the input is allowed to be "empty" (size=0)
if self.n_input != 0:
# reshape the data
u = inputData.reshape(self.n_input, 1)
outputData = outputData.reshape(self.n_output, 1)
# update the states
transmission = self.calculateLinearNetworkTransmissions(u, x)
x *= 1.0 - self._leakingRate
x += self._leakingRate * self._activation(transmission + B.dot(
self._WFeedback,
B.vstack((B.array(self._outputBias), outputData)),
) + (np.random.rand() - 0.5) * self._noiseLevel)
return u
else:
# reshape the data
outputData = outputData.reshape(self.n_output, 1)
# update the states
transmission = B.dot(self._W, x)
x *= 1.0 - self._leakingRate
x += self._leakingRate * self._activation(transmission + B.dot(
self._WFeedback,
B.vstack((B.array(self._outputBias), outputData)),
) + (np.random.rand() - 0.5) * self._noiseLevel)
return B.empty((0, 1))
def __init__(self, inputShape, n_reservoir,
filterSize=1, stride=1, borderMode="mirror", nWorkers="auto",
spectralRadius=1.0, noiseLevel=0.0, inputScaling=None,
leakingRate=1.0, reservoirDensity=0.2, randomSeed=None, averageOutputWeights=True,
out_activation=lambda x: x, out_inverse_activation=lambda x: x,
weightGeneration='naive', bias=1.0, outputBias=1.0,
outputInputScaling=1.0, inputDensity=1.0, solver='pinv', regressionParameters={}, activation=B.tanh,
activationDerivation=lambda x: 1.0 / B.cosh(x) ** 2):
self._averageOutputWeights = averageOutputWeights
if averageOutputWeights and solver != "lsqr":
raise ValueError(
"`averageOutputWeights` can only be set to `True` when `solver` is set to `lsqr` (Ridge Regression)")
self._borderMode = borderMode
if not borderMode in ["mirror", "padding", "edge", "wrap"]:
raise ValueError(
"`borderMode` must be set to one of the following values: `mirror`, `padding`, `edge` or `wrap`.")
self._regressionParameters = regressionParameters
self._solver = solver
n_inputDimensions = len(inputShape)
if filterSize % 2 == 0:
raise ValueError("filterSize has to be an odd number (1, 3, 5, ...).")
self._filterSize = filterSize
self._filterWidth = int(np.floor(filterSize / 2))
self._stride = stride
self._n_input = int(np.power(np.ceil(filterSize / stride), n_inputDimensions))
self.n_inputDimensions = n_inputDimensions
self.inputShape = inputShape
if not self._averageOutputWeights:
self._WOuts = B.empty((np.prod(inputShape), 1, self._n_input + n_reservoir + 1))
self._WOut = None
else:
self._WOuts = None
self._WOut = B.zeros((1, self._n_input + n_reservoir + 1))
self._xs = B.empty((np.prod(inputShape), n_reservoir, 1))
if nWorkers == "auto":
self._nWorkers = np.max((cpu_count() - 1, 1))
else:
self._nWorkers = nWorkers
manager = Manager()
self.sharedNamespace = manager.Namespace()
if hasattr(self, "fitWorkerID") == False or self.parallelWorkerIDs is None:
self.parallelWorkerIDs = manager.Queue()
for i in range(self._nWorkers):
self.parallelWorkerIDs.put((i))
super(SpatioTemporalESN, self).__init__(n_input=self._n_input, n_reservoir=n_reservoir, n_output=1,
spectralRadius=spectralRadius,
noiseLevel=noiseLevel, inputScaling=inputScaling,
leakingRate=leakingRate, reservoirDensity=reservoirDensity,
randomSeed=randomSeed, out_activation=out_activation,
out_inverse_activation=out_inverse_activation,
weightGeneration=weightGeneration, bias=bias, outputBias=outputBias,
outputInputScaling=outputInputScaling,
inputDensity=inputDensity, activation=activation,
activationDerivation=activationDerivation)
"""
def fit(self,
inputData,
outputData,
transientTime="AutoReduce",
transientTimeCalculationEpsilon=1e-3,
transientTimeCalculationLength=20,
verbose=0):
#check the input data
if self.n_input != 0:
if len(inputData.shape) == 3 and len(outputData.shape) > 1:
#multiple time series are used with a shape (timeseries, time, dimension) -> (timeseries, time, dimension)
if inputData.shape[0] != outputData.shape[0]:
raise ValueError(
"Amount of input and output datasets is not equal - {0} != {1}"
.format(inputData.shape[0], outputData.shape[0]))
if inputData.shape[1] != outputData.shape[1]:
raise ValueError(
"Amount of input and output time steps is not equal - {0} != {1}"
.format(inputData.shape[1], outputData.shape[1]))
else:
if inputData.shape[0] != outputData.shape[0]:
raise ValueError(
"Amount of input and output time steps is not equal - {0} != {1}"
.format(inputData.shape[0], outputData.shape[0]))
else:
if inputData is not None:
raise ValueError(
"n_input has been set to zero. Therefore, the given inputData will not be used."
)
if inputData is not None:
inputData = B.array(inputData)
if outputData is not None:
outputData = B.array(outputData)
#reshape the input/output data to have the shape (timeseries, time, dimension)
if len(outputData.shape) <= 2:
outputData = outputData.reshape((1, -1, self.n_output))
if inputData is not None:
if len(inputData.shape) <= 2:
inputData = inputData.reshape((1, -1, self.n_input))
self.resetState()
# Automatic transient time calculations
if transientTime == "Auto":
transientTime = self.calculateTransientTime(
inputData[0], outputData[0], transientTimeCalculationEpsilon,
transientTimeCalculationLength)
if transientTime == "AutoReduce":
if (inputData is None
and outputData.shape[2] == 1) or inputData.shape[2] == 1:
transientTime = self.calculateTransientTime(
inputData[0], outputData[0],
transientTimeCalculationEpsilon,
transientTimeCalculationLength)
transientTime = self.reduceTransientTime(
inputData[0], outputData[0], transientTime)
else:
print(
"Transient time reduction is supported only for 1 dimensional input."
)
if inputData is not None:
partialLength = (inputData.shape[1] - transientTime)
totalLength = inputData.shape[0] * partialLength
timeseriesCount = inputData.shape[0]
elif outputData is not None:
partialLength = (outputData.shape[1] - transientTime)
totalLength = outputData.shape[0] * partialLength
timeseriesCount = outputData.shape[0]
else:
raise ValueError("Either input or output data must not to be None")
self._X = B.empty((1 + self.n_input + self.n_reservoir, totalLength))
if (verbose > 0):
bar = progressbar.ProgressBar(max_value=totalLength,
redirect_stdout=True,
poll_interval=0.0001)
bar.update(0)
for i in range(timeseriesCount):
if inputData is not None:
xx, yy = self.propagate(inputData[i], outputData[i],
transientTime, verbose - 1)
self._X[:, i * partialLength:(i + 1) * partialLength] = xx
else:
xx, yy = self.propagate(None, outputData[i], transientTime,
verbose - 1)
self._X[:, i * partialLength:(i + 1) * partialLength] = xx
if (verbose > 0):
bar.update(i)
if (verbose > 0):
bar.finish()
#define the target values
Y_target = B.empty((outputData.shape[2], totalLength))
for i in range(timeseriesCount):
Y_target[:, i * partialLength:(i + 1) *
partialLength] = self.out_inverse_activation(
outputData[i]).T[:, transientTime:]
if (self._solver == "pinv"):
self._WOut = B.dot(Y_target, B.pinv(self._X))
#calculate the training prediction now
train_prediction = self.out_activation((B.dot(self._WOut,
self._X)).T)
# elif (self._solver == "lsqr"):
# X_T = self._X.T
# self._WOut = B.dot(B.dot(Y_target, X_T),B.inv(B.dot(self._X,X_T) + self._regressionParameters[0]*B.identity(1+self.n_input+self.n_reservoir)))
# """
# #alternative representation of the equation
# Xt = X.T
# A = np.dot(X, Y_target.T)
# B = np.linalg.inv(np.dot(X, Xt) + regression_parameter*np.identity(1+self.n_input+self.n_reservoir))
# self._WOut = np.dot(B, A)
# self._WOut = self._WOut.T
# """
# #calculate the training prediction now
# train_prediction = self.out_activation(B.dot(self._WOut, self._X).T)
elif (self._solver in [
"sklearn_auto", "sklearn_lsqr", "sklearn_sag", "sklearn_svd"
]):
mode = self._solver[8:]
params = self._regressionParameters
params["solver"] = mode
self._ridgeSolver = Ridge(**params)
self._ridgeSolver.fit(self._X.T, Y_target.T)
#calculate the training prediction now
train_prediction = self.out_activation(
self._ridgeSolver.predict(self._X.T))
elif (self._solver in ["sklearn_svr", "sklearn_svc"]):
self._ridgeSolver = SVR(**self._regressionParameters)
self._ridgeSolver.fit(self._X.T, Y_target.T.ravel())
#calculate the training prediction now
train_prediction = self.out_activation(
self._ridgeSolver.predict(self._X.T))
#calculate the training error now
#flatten the outputData
outputData = outputData[:, transientTime:, :].reshape(totalLength, -1)
training_error = B.sqrt(B.mean((train_prediction - outputData)**2))
return training_error
def fit_loop(self,
inputData,
outputData1,
outputData2,
transientTime="AutoReduce",
transientTimeCalculationEpsilon=1e-3,
transientTimeCalculationLength=20,
verbose=0):
#check the input data
if self.n_input != 0:
if len(inputData.shape) == 3 and len(outputData1.shape) > 1:
#multiple time series are used with a shape (timeseries, time, dimension) -> (timeseries, time, dimension)
if inputData.shape[0] != outputData1.shape[0]:
raise ValueError(
"Amount of input and output datasets is not equal - {0} != {1}"
.format(inputData.shape[0], outputData1.shape[0]))
if inputData.shape[1] != outputData1.shape[1]:
raise ValueError(
"Amount of input and output time steps is not equal - {0} != {1}"
.format(inputData.shape[1], outputData1.shape[1]))
else:
if inputData.shape[0] != outputData1.shape[0]:
raise ValueError(
"Amount of input and output time steps is not equal - {0} != {1}"
.format(inputData.shape[0], outputData1.shape[0]))
else:
if inputData is not None:
raise ValueError(
"n_input has been set to zero. Therefore, the given inputData will not be used."
)
inputData = B.array(inputData)
outputData1 = B.array(outputData1)
outputData2 = B.array(outputData2)
self._transientTime = transientTime
self._esn1.resetState()
self._esn2.resetState()
total_length = inputData.shape[0]
print("total_length ", total_length)
if (verbose > 0):
bar = progressbar.ProgressBar(max_value=total_length,
redirect_stdout=True,
poll_interval=0.0001)
bar.update(0)
#should be named aggregated_y
aggregated_y1 = B.empty((outputData1.shape))
aggregated_y2 = B.empty((outputData2.shape))
# X1, _ = self._esn1.propagate(inputData=inputData, outputData=outputData1, transientTime=transientTime, verbose=verbose-1)
# self._esn1._X = X1
# Y_target = self.out_inverse_activation(outputData1).T
# self._esn1._WOut = B.dot(Y_target, B.pinv(X1))
# y1 = self.out_activation((B.dot(self._esn1._WOut, X1)).T)
# training_error1 = B.sqrt(B.mean((y1.reshape((total_length, self._n_output1))- outputData1)**2))
# training_error2 = 0
y2 = self.out_activation(B.dot(self._esn2._WOut, self._esn2._X).T)
for i in range((total_length)):
#input: input + output from layer 2
inputDataWithFeedback = B.zeros((self.n_input + self._n_output2))
inputDataWithFeedback[:self.n_input] = inputData[i, :]
inputDataWithFeedback[self.n_input:] = y2
#update models
X1, untrained_y1 = self._esn1.propagateOneStep(
inputData=inputDataWithFeedback,
outputData=outputData1[i],
step=i,
transientTime=transientTime,
verbose=verbose - 1,
learn=True)
#self._esn1._X = X1
#y after model update (ideally we would use y before model update)
#y1 = self.out_activation((B.dot(self._esn1._WOut, X1)).T)
aggregated_y1[i, :] = untrained_y1.reshape(self._n_output1)
#output from 1st layer and correct output
# input2 = B.vstack((y1.reshape(self._n_output1), outputData1[i]))
# x2, y2 = self._esn2.propagateOneStep(inputData=input2, outputData=outputData2[i], step=i, transientTime=transientTime, verbose=verbose-1, learn=True)
# Y_target2[i,:] = y2.reshape(self._n_output2)
if (verbose > 0):
bar.update(i)
if (verbose > 0):
bar.finish()
self._training_res = aggregated_y1
training_error1 = B.sqrt(B.mean((aggregated_y1 - outputData1)**2))
training_error2 = B.sqrt(B.mean((aggregated_y2 - outputData2)**2))
print("training errors")
print(training_error1)
print(training_error2)
return training_error1, training_error2
def __init__(self,
n_input,
n_output1,
n_output2,
n_reservoir,
spectralRadius=1.0,
noiseLevel=0.0,
inputScaling=None,
leakingRate=1.0,
feedbackScaling=1.0,
reservoirDensity=0.2,
randomSeed=None,
out_activation=lambda x: x,
out_inverse_activation=lambda x: x,
weightGeneration='naive',
bias=1.0,
outputBias=1.0,
outputInputScaling=1.0,
inputDensity=1.0,
solver='pinv',
regressionParameters={},
activation=B.tanh,
activationDerivation=lambda x: 1.0 / B.cosh(x)**2):
#probably not needed
super(StackedESN,
self).__init__(n_input=n_input,
n_reservoir=n_reservoir,
n_output=n_output1,
spectralRadius=spectralRadius,
noiseLevel=noiseLevel,
inputScaling=inputScaling,
leakingRate=leakingRate,
feedbackScaling=feedbackScaling,
reservoirDensity=reservoirDensity,
randomSeed=randomSeed,
feedback=False,
out_activation=out_activation,
out_inverse_activation=out_inverse_activation,
weightGeneration=weightGeneration,
bias=bias,
outputBias=outputBias,
outputInputScaling=outputInputScaling,
inputDensity=inputDensity,
activation=activation,
activationDerivation=activationDerivation)
#layers
self._esn1 = PredictionESN(
n_input=(n_input + n_output2),
n_reservoir=n_reservoir,
n_output=n_output1,
spectralRadius=spectralRadius,
noiseLevel=noiseLevel,
inputScaling=inputScaling,
leakingRate=leakingRate,
feedbackScaling=feedbackScaling,
reservoirDensity=reservoirDensity,
randomSeed=randomSeed,
feedback=False,
out_activation=out_activation,
out_inverse_activation=out_inverse_activation,
weightGeneration=weightGeneration,
bias=bias,
outputBias=outputBias,
outputInputScaling=outputInputScaling,
inputDensity=inputDensity,
activation=activation,
activationDerivation=activationDerivation)
#esn_2 takes coordinates in (estimated and actual)
self._esn2 = PredictionESN(
n_input=n_output1 * 2,
n_reservoir=n_reservoir,
n_output=n_output2,
spectralRadius=spectralRadius,
noiseLevel=noiseLevel,
inputScaling=inputScaling,
leakingRate=leakingRate,
feedbackScaling=feedbackScaling,
reservoirDensity=reservoirDensity,
randomSeed=randomSeed,
feedback=False,
out_activation=out_activation,
out_inverse_activation=out_inverse_activation,
weightGeneration=weightGeneration,
bias=bias,
outputBias=outputBias,
outputInputScaling=outputInputScaling,
inputDensity=inputDensity,
activation=activation,
activationDerivation=activationDerivation)
self._solver = solver
self._regressionParameters = regressionParameters
self._transientTime = 0
self._n_output1 = n_output1
self._n_output2 = n_output2
self._training_res = B.empty((2, 2))
"""
def fit(self,
inputData,
outputData1,
outputData2,
transientTime="AutoReduce",
transientTimeCalculationEpsilon=1e-3,
transientTimeCalculationLength=20,
verbose=0):
#check the input data
if self.n_input != 0:
if len(inputData.shape) == 3 and len(outputData1.shape) > 1:
#multiple time series are used with a shape (timeseries, time, dimension) -> (timeseries, time, dimension)
if inputData.shape[0] != outputData1.shape[0]:
raise ValueError(
"Amount of input and output datasets is not equal - {0} != {1}"
.format(inputData.shape[0], outputData1.shape[0]))
if inputData.shape[1] != outputData1.shape[1]:
raise ValueError(
"Amount of input and output time steps is not equal - {0} != {1}"
.format(inputData.shape[1], outputData1.shape[1]))
else:
if inputData.shape[0] != outputData1.shape[0]:
raise ValueError(
"Amount of input and output time steps is not equal - {0} != {1}"
.format(inputData.shape[0], outputData1.shape[0]))
else:
if inputData is not None:
raise ValueError(
"n_input has been set to zero. Therefore, the given inputData will not be used."
)
inputData = B.array(inputData)
outputData1 = B.array(outputData1)
outputData2 = B.array(outputData2)
self._transientTime = transientTime
self._esn1.resetState()
self._esn2.resetState()
total_length = inputData.shape[0]
print("total_length ", total_length)
aggregated_y1 = B.empty((outputData1.shape))
aggregated_y2 = B.empty((outputData2.shape))
#put pixels and switch data together
inputDataWithFeedback = B.zeros(
(total_length, self.n_input + self._n_output2))
inputDataWithFeedback[:, :self.n_input] = inputData
inputDataWithFeedback[:, self.n_input:] = outputData2
X1, _ = self._esn1.propagate(inputData=inputDataWithFeedback,
outputData=outputData1,
transientTime=transientTime,
verbose=verbose - 1)
self._esn1._X = X1
Y_target = self.out_inverse_activation(outputData1).T
self._esn1._WOut = B.dot(Y_target, B.pinv(X1))
aggregated_y1 = self.out_activation((B.dot(self._esn1._WOut, X1)).T)
training_error1 = B.sqrt(
B.mean((aggregated_y1.reshape(
(total_length, self._n_output1)) - outputData1)**2))
training_error2 = 0
training_error1 = B.sqrt(B.mean((aggregated_y1 - outputData1)**2))
training_error2 = B.sqrt(B.mean((aggregated_y2 - outputData2)**2))
return training_error1, training_error2
def reduceTransientTime(self, inputs, outputs, initialTransientTime, epsilon = 1e-3, proximityLength = 50):
# inputs: input of reserovoir
# outputs: output of reservoir
# epsilon: given constant
# proximity length: number of steps for which all states have to be epsilon close to declare convergance
# initialTransientTime: transient time with calculateTransientTime() method estimated
# finds initial state with lower transient time and sets internal state to this state
# returns the new transient time by calculating the convergence time of initial states found with SWD and Equilibrium method
def getEquilibriumState(inputs, outputs, epsilon = 1e-3):
# inputs: input of reserovoir
# outputs: output of reservoir
# epsilon: given constant
# returns the equilibrium state when esn is fed with the first state of input
x = B.empty((2, self.n_reservoir, 1))
while not B.max(B.ptp(x, axis=0)) < epsilon:
x[0] = x[1]
u = inputs[0].reshape(-1, 1) if inputs is not None else None
o = outputs[0].reshape(-1, 1) if outputs is not None else None
self.update(u, o, x[1])
return x[1]
def getStateAtGivenPoint(inputs, outputs, targetTime):
# inputs: input of reserovoir
# outputs: output of reservoir
# targetTime: time at which the state is wanted
# propagates the inputs/outputs till given point in time and returns the state of the reservoir at this point
x = B.zeros((self.n_reservoir, 1))
length = inputs.shape[0] if inputs is not None else outputs.shape[0]
length = min(length, targetTime)
for t in range(length):
u = inputs[t].reshape(-1, 1) if inputs is not None else None
o = outputs[t].reshape(-1, 1) if outputs is not None else None
self.update(u, o, x)
return x
length = inputs.shape[0] if inputs is not None else outputs.shape[0]
if proximityLength is None:
proximityLength = int(length * 0.1)
if proximityLength < 3:
proximityLength = 3
x = B.empty((2, self.n_reservoir, 1))
equilibriumState = getEquilibriumState(inputs, outputs)
if inputs is None:
swdPoint, _ = hp.SWD(outputs, int(initialTransientTime*0.8))
else:
swdPoint, _ = hp.SWD(inputs, int(initialTransientTime * 0.8))
swdState = getStateAtGivenPoint(inputs, outputs, swdPoint)
x[0] = equilibriumState
x[1] = swdState
transientTime = 0
countedConsecutiveSteps = 0
for t in range(length):
if B.max(B.ptp(x, axis=0)) < epsilon:
countedConsecutiveSteps += 1
if countedConsecutiveSteps > proximityLength:
transientTime = t - proximityLength
break
else:
countedConsecutiveSteps = 0
u = inputs[t].reshape(-1, 1) if inputs is not None else None
o = outputs[t].reshape(-1, 1) if outputs is not None else None
for i in range(x.shape[0]):
self.update(u, o, x[i])
self._x = x[0]
return transientTime
def propagate(self, inputData, outputData=None, transientTime=0, verbose=0, x=None, steps="auto", feedbackData=None):
if x is None:
x = self._x
inputLength = steps
if inputData is None:
if outputData is not None:
inputLength = len(outputData)
else:
inputLength = len(inputData)
if inputLength == "auto":
raise ValueError("inputData and outputData are both None. Therefore, steps must not be `auto`.")
# define states' matrix
X = B.zeros((1 + self.n_input + self.n_reservoir, inputLength - transientTime))
if (verbose > 0):
bar = progressbar.ProgressBar(max_value=inputLength, redirect_stdout=True, poll_interval=0.0001)
bar.update(0)
if self._WFeedback is None:
#do not distinguish between whether inputData is None or not, as the feedback has been disabled
#therefore, the input has to be anything but None
for t in range(inputLength):
u, x = self.update(inputData[t], x=x)
if (t >= transientTime):
#add valueset to the states' matrix
X[:,t-transientTime] = B.vstack((B.array(self._outputBias), self._outputInputScaling*u, x))[:,0]
Y = B.dot(self._WOut, X)
if (verbose > 0):
bar.update(t)
else:
if outputData is None:
Y = B.empty((inputLength-transientTime, self.n_output))
if feedbackData is None:
feedbackData = B.zeros((1, self.n_output))
if inputData is None:
for t in range(inputLength):
self.update(None, feedbackData, x=x)
if (t >= transientTime):
#add valueset to the states' matrix
X[:,t-transientTime] = B.vstack((B.array(self._outputBias), x))[:,0]
if outputData is None:
#calculate the prediction using the trained model
if (self._solver in ["sklearn_auto", "sklearn_lsqr", "sklearn_sag", "sklearn_svd"]):
feedbackData = self._ridgeSolver.predict(B.vstack((B.array(self._outputBias), self._x)).T)
else:
feedbackData = B.dot(self._WOut, B.vstack((B.array(self._outputBias), self._x)))
if t >= transientTime:
Y[t-transientTime, :] = feedbackData
else:
feedbackData = outputData[t]
if (verbose > 0):
bar.update(t)
else:
for t in range(inputLength):
u = self.update(inputData[t], feedbackData, x=x)
if (t >= transientTime):
#add valueset to the states' matrix
X[:,t-transientTime] = B.vstack((B.array(self._outputBias), self._outputInputScaling*u, x))[:,0]
if outputData is None:
#calculate the prediction using the trained model
if (self._solver in ["sklearn_auto", "sklearn_lsqr", "sklearn_sag", "sklearn_svd"]):
feedbackData = self._ridgeSolver.predict(B.vstack((B.array(self._outputBias), self._outputInputScaling*u, self._x)).T)
else:
feedbackData = B.dot(self._WOut, B.vstack((B.array(self._outputBias), self._outputInputScaling*u, self._x)))
Y[t, :] = feedbackData.ravel()
else:
feedbackData = outputData[t]
if (verbose > 0):
bar.update(t)
if (verbose > 0):
bar.finish()
return X, Y
def __init__(
self,
inputShape,
n_reservoir,
filterSize=1,
stride=1,
borderMode="mirror",
nWorkers="auto",
spectralRadius=1.0,
noiseLevel=0.0,
inputScaling=None,
leakingRate=1.0,
reservoirDensity=0.2,
randomSeed=None,
averageOutputWeights=True,
out_activation=lambda x: x,
out_inverse_activation=lambda x: x,
weightGeneration="naive",
bias=1.0,
outputBias=1.0,
outputInputScaling=1.0,
inputDensity=1.0,
solver="pinv",
regressionParameters={},
activation=B.tanh,
activationDerivative=lambda x: 1.0 / B.cosh(x)**2,
chunkSize=16,
):
""" ESN that predicts (steps of) a spatio-temporal time series based on a time series.
Args:
inputShape : Shape of the input w/o the time axis, e.g. (W, H) for a 2D input.
n_reservoir : Number of units in the reservoir.
filterSize : Size of patches used to predict a single output element.
stride : Stride between different patches.
borderMode : How to handle border values. Choices: mirror, padding, edge, wrap.
nWorkers : Number of CPU threads executed in parallel to solve the problem.
spectralRadius : Spectral radius of the reservoir's connection/weight matrix.
noiseLevel : Magnitude of noise that is added to the input while fitting to prevent overfitting.
inputScaling : Scaling factor of the input.
leakingRate : Convex combination factor between 0 and 1 that weights current and new state value.
reservoirDensity : Percentage of non-zero weight connections in the reservoir.
randomSeed : Seed for random processes, e.g. weight initialization.
averageOutputWeights : Average output matrices after fitting across all pixels or use a distinct matrix
per pixel. The former assumes homogeneity of the problem across all pixels.
out_activation : Final activation function (i.e. activation function of the output).
out_inverse_activation : Inverse of the final activation function
weightGeneration : Algorithm to generate weight matrices. Choices: naive, SORM, advanced, custom
bias : Size of the bias added for the internal update process.
outputBias : Size of the bias added for the final linear regression of the output.
outputInputScaling : Rescaling factor for the input of the ESN for the regression.
inputDensity : Percentage of non-zero weights in the input-to-reservoir weight matrix.
solver : Algorithm to find output matrix. Choices: pinv, lsqr.
regressionParameters : Arguments to the solving algorithm. For LSQR this controls the L2 regularization.
activation : (Non-linear) Activation function.
activationDerivative : Derivative of the activation function.
chunkSize : Internal parameter for the multi-threading. For long time series this should be reduced to
avoid OOM errors/getting stuck and to reduce memory consumption.
"""
self._averageOutputWeights = averageOutputWeights
if averageOutputWeights and solver != "lsqr":
raise ValueError(
"`averageOutputWeights` can only be set to `True` when `solver` is set to `lsqr` (Ridge Regression)"
)
self._borderMode = borderMode
if not borderMode in ["mirror", "padding", "edge", "wrap"]:
raise ValueError(
"`borderMode` must be set to one of the following values: `mirror`, `padding`, `edge` or `wrap`."
)
self._regressionParameters = regressionParameters
self._solver = solver
n_inputDimensions = len(inputShape)
if filterSize % 2 == 0:
raise ValueError(
"filterSize has to be an odd number (1, 3, 5, ...).")
self._filterSize = filterSize
self._filterWidth = int(np.floor(filterSize / 2))
self._stride = stride
self._n_input = int(
np.power(np.ceil(filterSize / stride), n_inputDimensions))
self.n_inputDimensions = n_inputDimensions
self.inputShape = inputShape
if not self._averageOutputWeights:
self._WOuts = B.empty(
(np.prod(inputShape), 1, self._n_input + n_reservoir + 1))
self._WOut = None
else:
self._WOuts = None
self._WOut = B.zeros((1, self._n_input + n_reservoir + 1))
self._xs = B.empty((np.prod(inputShape), n_reservoir, 1))
if nWorkers == "auto":
self._nWorkers = np.max((cpu_count() - 1, 1))
else:
self._nWorkers = nWorkers
manager = Manager()
self.sharedNamespace = manager.Namespace()
if hasattr(self,
"fitWorkerID") == False or self.parallelWorkerIDs is None:
self.parallelWorkerIDs = manager.Queue()
for i in range(self._nWorkers):
self.parallelWorkerIDs.put((i))
self._chunkSize = chunkSize
super(SpatioTemporalESN, self).__init__(
n_input=self._n_input,
n_reservoir=n_reservoir,
n_output=1,
spectralRadius=spectralRadius,
noiseLevel=noiseLevel,
inputScaling=inputScaling,
leakingRate=leakingRate,
reservoirDensity=reservoirDensity,
randomSeed=randomSeed,
out_activation=out_activation,
out_inverse_activation=out_inverse_activation,
weightGeneration=weightGeneration,
bias=bias,
outputBias=outputBias,
outputInputScaling=outputInputScaling,
inputDensity=inputDensity,
activation=activation,
activationDerivative=activationDerivative,
)
"""