The ability to quantify the uncertainty in our models of nature is fundamental to many inference problems in science and engineering. In this course, Ravela will lead a study of advanced methods to represent, sample, update and propagate uncertainty.
This is a "hands on" course: Methodology will be coupled with applications. The course will include lectures, invited talks, discussions, reviews and projects and will meet once a week to discuss a method and its applications.
Methods under consideration will include Markov Chain Monte Carlo, Polynomial Chaos, Spectral and Scale-space Model Reduction, Expectation Maximization, Hierarchical and Variational Bayes, Particle Filter and Smoother, Graphical Models, Spatial Inference, Principal Modes - EOF (PCA), ICA, NNMF, UCA..., Reconstruction and Inverse Problems with L1, L2... L-Infinity or mutual information measures, under entropy, smoothness, sparsity or compressibility.
Applications to include: Autonomous Observation, Climate Reconstruction, Data Assimilation/ State Estimation, Image Processing, Natural Hazards and Predictability.
Instructor: Sai Ravela
First class: Friday, February 10th in 54-1623 @10am.
(Full Photo Credit: http://upload.wikipedia.org/wikipedia/commons/e/e0/Lorenz.png)