Speaker: Yuri Kalnishkan (Royal Holloway University of London) 

Title: Predictive Complexity 

Abstract:
The concept of predictive complexity is a generalisation of
Kolmogorov complexity motivated by the computational learning theory.
The theory of predictive complexity is a development of prediction
with expert advice. The predictive complexity of a sequence is an
inherent measure of how difficult it is to predict the elements of the
sequence.
The talk discusses some results about predictive complexity:
existence of predictive complexity and weak predictive complexities
(what loss functions correspond to predictive complexities),
expectations and linear inequalities (how to compare predictability
under different loss functions), uniqueness problem (predictive
complexity specifies a unique loss function up to a parametrisation)
unpredictability property (most sequence are unpredictable in some
precise sense).

Bio:
Graduated from the Department of Mathematics and Mechanics, Moscow
State University, 1998
PhD Royal Holloway, University of London, 2002
Have been a lecturer at the Department of Computer Science, Royal
Holloway, University of London since 2003

URL: http://www.clrc.rhul.ac.uk/people/yura/