def run(self,input):
data = input["hvlog"]
max_ind = numpy.max(input["mlog"])
data = data[0:max_ind,:]
data = data.transpose()
N,M = data.shape
H = numpy.max(numpy.array([2,numpy.ceil(M/3)]))
T = 50
maxIter = 20
A = numpy.eye(H, H)
C = numpy.eye(M, H)
Q = numpy.eye(H, H)
R = numpy.eye(M, M)
mu0 = numpy.random.randn(1,H)
Q0 = Q
model = kalman.lds_model(A,C,Q,R,mu0,Q0)
z_hat,y_hat,ind = kalman.learn_lds(model,data,T,H,self.step_size,maxIter,self.lookahead_flag)
output = input
output["predict"] = y_hat.transpose()
output["smooth"] = z_hat.transpose()
output["ind"] = np.array(ind)
return input
def run(self, input):
data = input["hvlog"]
max_ind = numpy.max(input["mlog"])
data = data[0:max_ind, :]
data = data.transpose()
N, M = data.shape
H = numpy.max(numpy.array([2, numpy.ceil(M / 3)]))
T = 50
maxIter = 20
A = numpy.eye(H, H)
C = numpy.eye(M, H)
Q = numpy.eye(H, H)
R = numpy.eye(M, M)
mu0 = numpy.random.randn(1, H)
Q0 = Q
model = kalman.lds_model(A, C, Q, R, mu0, Q0)
z_hat, y_hat, ind = kalman.learn_lds(model, data, T, H, self.step_size,
maxIter, self.lookahead_flag)
output = input
output["predict"] = y_hat.transpose()
output["smooth"] = z_hat.transpose()
output["ind"] = np.array(ind)
return input
## Model Initialization
N,M = y_noise.shape
# A = np.array([[1,0,1,0],[0,1,0,1],[0,0,1,0],[0,0,0,1]])
# C = np.array([[1,0,0,0],[0,1,0,0]])
# Q = 0.1*np.eye(H)
# R = 0.1*np.eye(M)
# mu0 = np.array([10,10,1,0])
A = 0.1*np.eye(H, H) + 0*np.random.randn(H, H)
C = 0.1*np.eye(M, H) + 0*np.random.randn(M, H)
Q = np.eye(H, H)
R = np.eye(M, M)
mu0 = np.random.randn(1,H)
Q0 = Q
model = kalman.lds_model(A,C,Q,R,mu0,Q0)
z_hat,y_hat = kalman.learn_lds(model,y_noise,T,H,step_size,maxIter)
plt.figure(1)
plt.scatter(y_noise[:,0],y_noise[:,1])
plt.scatter(y_hat[:,0],y_hat[:,1],c='r')
plt.title('Predicted vs Actual Location')
plt.figure(2)
plt.plot(np.sqrt(np.sum((y_hat-y_noise)**2,1)))
plt.title('Location Error')
plt.show()
## Model Initialization
N, M = y_noise.shape
# A = np.array([[1,0,1,0],[0,1,0,1],[0,0,1,0],[0,0,0,1]])
# C = np.array([[1,0,0,0],[0,1,0,0]])
# Q = 0.1*np.eye(H)
# R = 0.1*np.eye(M)
# mu0 = np.array([10,10,1,0])
A = 0.1 * np.eye(H, H) + 0 * np.random.randn(H, H)
C = 0.1 * np.eye(M, H) + 0 * np.random.randn(M, H)
Q = np.eye(H, H)
R = np.eye(M, M)
mu0 = np.random.randn(1, H)
Q0 = Q
model = kalman.lds_model(A, C, Q, R, mu0, Q0)
z_hat, y_hat = kalman.learn_lds(model, y_noise, T, H, step_size, maxIter)
plt.figure(1)
plt.scatter(y_noise[:, 0], y_noise[:, 1])
plt.scatter(y_hat[:, 0], y_hat[:, 1], c='r')
plt.title('Predicted vs Actual Location')
plt.figure(2)
plt.plot(np.sqrt(np.sum((y_hat - y_noise)**2, 1)))
plt.title('Location Error')
plt.show()
