SOME CONTEMPORARY STATISTICAL INVERSE PROBLEMS
by Professor Peter Hall
Abstract: As technology advances, it alters not only the tools available for
statistical analysis, but also the type of data that need to be analysed.
Therefore statistical methodology is under pressure from at least two
sides: New techniques need to be introduced to exploit newly available
levels of computational power, and new methodologies are needed to handle
new types of data, and respond to questions that have new, practical
motivations. Illustrating these directions, we shall address several
nonstandard statistical `inversion' problems, including hyperspectral
target detection and the unfolding of evolving periodic functions.
For More Information: Department of Mathematics and Statistics: Ph. 8344 5550