[YPNG] YPNG Friday 12 September 2014, 11:00-1:00, usual place

[sekhar]

Hi Everyone,

Patrick Rebeschini will talk about  how to tweak classical sequential 
Monte Carlo algorithms
(particle filters) to overcome the curse of dimensionality in high 
dimensional problems. He
will review the classical theory (bootstrap particle filter), analyze 
the curse of dimensionality,
and discuss what ideas are needed to address the core of the problem and 
to make tangible
progress in applications.

See you Friday at 11 in the Stat's classroom.

Regards,
sekhar


On 8/12/2014 3:12 PM, David Pollard wrote:
> Folks,
>
> I'll tell you about an interested example of a Gaussian process that illustrates nicely
> the difference between two techniques for controlling the expected supremum
> of a Gaussian process. The example is an improved version of an classical example that took
> over a YPNG seminar in January 2013.  The audience suggestions have subsequently
> proved most helpful.
>
> For anyone who feels the need to prepare for the seminar, see section 4 of:
> https://dl.dropboxusercontent.com/u/35268723/Packing.pdf
> _______________________________________________
> YPNG mailing list
> YPNG at mailman.yale.edu
> http://mailman.yale.edu/mailman/listinfo/ypng