On using fishers information matrix to build a new general theory for dimension reduction by linear transformation Seminar
- Date:
- 31 January 2013
- Venue:
- Building 58 Room 1009
For more information regarding this seminar, please email Mrs Jane Revell at j.revell@southamton.ac.uk .
Event details
Special lecture
It has long been a standard practice to reduce the dimension of data by choosing a small subset of linear combinations of the data (often called projections). We will show that the Fisher's information matrix, appropriately defined, can be used as a very general tool to find useful projections of data.
Among the multivariate methodologies that the method touches upon are projection pursuit, discrimination analysis, independent components analysis, and the reduction of dimension in regression variables.
Speaker information
Buce Lindsay , Pennsylvania State University. Bruce Lindsay is the holder of the Eberly Chair in Statistics (previously hold by C.R. Rao) at the Department of Statistics, The Pennsylvania State University in the US. Bruce has done major contributions to the development of statistics including important characterizations in the geometry of mixture likelihoods, algorithms, likelihood inference with nuisances parameters and biometrical applications. Bruce has produced an enormous number of PhD students and a large number of highest quality publications. His talks are always full of inspiration.