Reconstruction of NDVI Data with Normal VarianceMean Mixture Noise using Stochastic Gradient Methods
(2017) FMS820 20171Mathematical Statistics
 Abstract
 Satellite data is useful for making inference on large areas, but there are plenty of problems associated with the technology. The observation errors need to be carefully modelled in order to make inference from the data. In addition, the large datasets involved makes straightforward computations infeasible.
This thesis explores a nonGaussian noise structure, and implements a modiﬁed EM algorithm where the Mstep is replaced for taking a step in a stochastic gradient descent method. The estimated parameters are used for reconstruction both of simulated data and of a dataset retrieved from the PAL database.
The algorithm performs well on a simulated data equipped with the nonGaussian noise structure; as for the PAL data, the... (More)  Satellite data is useful for making inference on large areas, but there are plenty of problems associated with the technology. The observation errors need to be carefully modelled in order to make inference from the data. In addition, the large datasets involved makes straightforward computations infeasible.
This thesis explores a nonGaussian noise structure, and implements a modiﬁed EM algorithm where the Mstep is replaced for taking a step in a stochastic gradient descent method. The estimated parameters are used for reconstruction both of simulated data and of a dataset retrieved from the PAL database.
The algorithm performs well on a simulated data equipped with the nonGaussian noise structure; as for the PAL data, the algorithm performed quite well, but convergence proved to be somewhat unstable. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/studentpapers/record/8902621
 author
 Kraft, David
 supervisor

 Johan Lindström ^{LU}
 organization
 course
 FMS820 20171
 year
 2017
 type
 H2  Master's Degree (Two Years)
 subject
 language
 English
 id
 8902621
 date added to LUP
 20170207 08:19:48
 date last changed
 20170207 08:19:48
@misc{8902621, abstract = {Satellite data is useful for making inference on large areas, but there are plenty of problems associated with the technology. The observation errors need to be carefully modelled in order to make inference from the data. In addition, the large datasets involved makes straightforward computations infeasible. This thesis explores a nonGaussian noise structure, and implements a modiﬁed EM algorithm where the Mstep is replaced for taking a step in a stochastic gradient descent method. The estimated parameters are used for reconstruction both of simulated data and of a dataset retrieved from the PAL database. The algorithm performs well on a simulated data equipped with the nonGaussian noise structure; as for the PAL data, the algorithm performed quite well, but convergence proved to be somewhat unstable.}, author = {Kraft, David}, language = {eng}, note = {Student Paper}, title = {Reconstruction of NDVI Data with Normal VarianceMean Mixture Noise using Stochastic Gradient Methods}, year = {2017}, }