MPI-INF Logo
Campus Event Calendar

Event Entry

New for: D1, D2, D3, D4, D5

What and Who

Epsilon ID's and the double sampling argument

Shay Moran
Max-Planck-Institut für Informatik - D1
Lecture
AG 1, AG 2, AG 3, AG 4, AG 5, RG1, SWS, MMCI  
AG Audience
English

Date, Time and Location

Wednesday, 15 January 2014
13:00
60 Minutes
E1 4
024
Saarbrücken

Abstract

I will introduce the concept of Epsilon ID's (a.k.a Epsilon-nets)

and discuss the fundamental connection of it
with PAC learnability.

This connection is usually established via the
Vapnik-Chervonenkis dimension (VC dimension).
However, in this talk we will abstract away the definition
of the VC dimension and we will show that instead of the VC dimension
one can use any complexity measure
which is defined on sets of boolean functions and possesses
two, rather natural, properties.

The lecture will be self-contained and will last roughly 1 hour.

Contact

Shay Moran
--email hidden
passcode not visible
logged in users only

Shay Moran, 07/15/2014 08:43 -- Created document.