The University of Maine
  Calendar  |  Campus Map  | 
About UMaine | Student Resources | Prospective Students
Faculty & Staff
| Alumni | Arts | News | Parents | Research


Physics and
Astronomy

Links

division
 Undergraduate
 Programs

division
 Graduate Programs

division
 Research
division
 People
division
 Student Opportunities
division
 Public Services
division
 Department Timeline
division
 Sitemap
division
 Physics Organizations

division
 Location

division
 Mainely Physics

division
 Newsworthy Notes
division
 Physics Colloquium

division
 Links of Interest

division
 
 Related Programs

division
 LASST

division
 Jordan Planetarium

division
 Center for Science
 and Mathematics
 Education Research

division
 MST Program

division




Physics and Astronomy


Physics Colloquium - Spring 2007 - Parameters of stochastic diffusion processes estimated from observations of first hitting-times: application to the leaky integrate-and-fire neuronal model

Dept of Physics & Astronomy
University of Maine, Orono, Maine

Presents

Susanne Ditlevsen
Associate Professor
Dept. of Biostatistics, University of Copenhagen

Parameters of stochastic diffusion processes estimated from observations of first hitting-times: application to the leaky integrate-and-fire neuronal model

The first hitting-time to a constant threshold of a diffusion process has been in focus for stochastic modeling of problems where a hidden random process only shows when it reaches a certain level that triggers some observable event. Applications come from various fields, e.g. neuronal modeling, survival analysis, mathematical finance, and more generally from physics, biology, finance, engineering and many others.

Apart from studying properties of the models themselves, methods for model verification are equally important. The first step is to identify model parameters estimated from experimental data, then statistical comparison between the model and the data can be performed. In many applications where renewal point data are available, models of first hitting-times of underlying diffusion processes arise. Despite of the seemingly simplicity of the model, the problem of how to estimate parameters of the underlying stochastic process has resisted its solution. The few attempts have either been unreliable, difficult to implement or only valid in subsets of the relevant parameter space. In this talk a newly developed estimation method that overcomes these difficulties is presented, it is computationally easy and fast to implement, and also works surprisingly well on small data sets. It is a direct application of the Fortet integral equation. The method is illustrated on simulated data and applied to recordings of neuronal activity. It also provides a diagnostic tool, which in the neuronal setting can distinguish between bursting and non-bursting behavior.

Friday,  February 16, 2007

3:10 pm

140 Bennett Hall

Refreshments will follow in Rm. 114, Bennett Hall


Back to Physics Colloquium - Spring 2007

 

Department of Physics
120 Bennett Hall
Orono, Maine 04469-5709
Phone: (207) 581-1039 | Fax: (207) 581-3410
Chairperson: Dr. David Batuski


The University of Maine
, Orono, Maine 04469
207-581-1110
A Member of the University of Maine System