Connectionists: Call for participation: NIPS workshop on On-line Trading of Exploration and Exploitation

[Zakria Hussain]

CALL FOR PARTICIPATION:

----------------------------------------------------------------------------
-----------------

	On-line Trading of Exploration and Exploitation
	http://www.homepages.ucl.ac.uk/~ucabzhu/OTEE.htm
	
		Nips 2006 Workshop
	December 8-9th, Whistler, BC, Canada
----------------------------------------------------------------------------
------------------

Background:

Trading exploration and exploitation plays a key role in a number of
learning tasks. For example the bandit problem ([1],[2],[3],[4]) provides
perhaps the simplest case in which we must decide a trade-off between
pulling the arm that appears most advantageous and experimenting with arms
for which we do not have accurate information. Similar issues arise in
learning problems where the information received depends on the choices made
by the learner. Examples include reinforcement learning and active learning,
though similar issues also arise in other disciplines, for example
sequential decision-making from statistics, optimal control from control
theory, etc.

Learning studies have frequently concentrated on the final performance of
the learned system rather than consider the errors made during the learning
process. For example reinforcement learning has traditionally been concerned
with showing convergence to an optimal policy, while in contrast analysis of
the bandit problem has attempted to bound the extra loss experienced during
the learning process when compared with an a priori optimal agent.

The workshop will provide a focus for work concerned with on-line trading of
exploration and exploitation, in particular providing a forum for extensions
to the bandit problem, invited presentations by researchers working in
related areas in other disciplines, as well as discussion and contributed
papers.


Call for Participation:

Papers: The organizing committee would like to invite extended abstract
paper submissions to the NIPS 2006 workshop on ‘On-line Trading of
Exploration and Exploitation’ in the following related areas (but not
restricted to):

    * Exploration and Exploitation problems
    * Multi-armed bandit problems
    * Sequential decision-making
    * Empirical/theoretical studies of bandit problems
    * On-line learning algorithms
    * Related work from other disciplines such as control theory, game
theory, statistics etc.

The organizers will select a small number of contributed papers for oral
presentation to last between 20 - 40 minutes. All other accepted papers will
be displayed in a poster session. We shall evaluate proposed talks on the
merits of both their relevance and significance and their contribution to
the well-roundedness of the workshop as a whole.

Paper submission deadline: 28th October 2006
Acceptance notification: 10th November 2006
Workshop: 8th or 9th December

Papers should be written using the NIPS style file (nips06.sty), not exceed
8 pages in length, and be sent as PDF files to Zakria Hussain by 23.59
(Western Samoa time) on the 28th October 2006.

Challenge: A call for participation in phase 2 of the PASCAL network of
excellence Exploration Vs Exploitation (EE) challenge (see
http://www.pascal-network.org/Challenges/EEC/). The workshop will be used to
discuss the results of the challenge currently running under the auspices of
the PASCAL network and Touch Clarity Ltd, investigating algorithms for
multi-armed bandits in which the response rates for the individual arms vary
over time according to certain patterns. Touch Clarity have agreed to award
a first prize of £1000 to the best entry. See challenge website for
instructions, data set downloads and submission instructions.

Code submission: 15th November 2006
Seed distribution: 16th November 2006
Final results: 23rd November 2006

Structure of the workshop:

We plan a 1-day workshop to be held either on the 8th or 9th December 2006.

The workshop will include a discussion of the results of the PASCAL
challenge concerned with extensions of the bandit problem as well as
broadening the discussion to other tasks and other approaches to on-line
trading of exploration and exploitation.

Our aim is to make connections to approaches (both algorithmic and
theoretical) that have been developed in other areas. We plan to fund
invited speakers from other disciplines, specifically a presentation on
Optimal Control and on Sequential Decision-making.

One part of the workshop will be used to discuss the results of the
challenge (see above) currently running under the auspices of the PASCAL
network investigating algorithms for multi-armed bandits in which the
response rates for the individual arms vary over time according to certain
patterns (see http://www.pascal-network.org/Challenges/EEC/).


An outline schedule:

    * Presentation of the practical problem underlying the PASCAL challenge
    * Short spotlight presentations of successful approaches to the
challenge interspersed with discussion leading into a general discussion of
results and lessons learnt.
    * Invited talks/tutorials on other approaches in related disciplines
specifically optimal control and sequential decision-making. A Preliminary
list of invited speakers is: Adam Kalai and Csaba Szepesvári (to be
confirmed).
    * Discussion of relations to more standard learning formulations such as
reinforcement learning.
    * Contributed talks interspersed with discussion
    * Poster session
    * Closing discussion

Program Committee:

Peter Auer 		University of Leoben
Nicolò Cesa-Bianchi 	University of Milan
Zakria Hussain 		University College London
Adam Kalai 		Toyota Technology Institute (to be confirmed)
Robert Kleinberg 	University of California Berkeley (to be confirmed)
Yishay Mansour 	Tel Aviv University
Leonard Newnham 	Touch Clarity Ltd
John Shawe-Taylor 	University College London

Organization:

Peter Auer 		University of Leoben
Nicolò Cesa-Bianchi 	University of Milan
Zakria Hussain 		University College London
Leonard Newnham 	Touch Clarity Ltd
John Shawe-Taylor 	University College London
	
References:

[1] Peter Auer, Nicolò Cesa-Bianchi, Paul Fischer. Finite-time Analysis of
the Multi-armed Bandit Problem. Machine Learning (2002).

[2] Peter Auer, Nicolò Cesa-Bianchi, Yoav Freund, Robert E. Schapire.
Gambling in a Rigged Casino: The adversarial multi-armed bandit problem.
Proceedings of the 36th Annual Symposium on Foundations of Computer Science
(1998).

[3]Shie Mannor, John N. Tsitsiklis. The Sample Complexity of Exploration in
the Multi-Armed Bandit Problem. The Journal of Machine Learning Research
(2004).

[4] Joannès Vermorel and Mehryar Mohri. Multi-Armed Bandit Algorithms and
Empirical Evaluation. In Proceedings of the 16th European Conference on
Machine Learning (ECML 2005).

Sponsors:
PASCAL network of excellence
Touch Clarity Ltd