| EPSRC Reference: |
EP/F025645/1 |
| Title: |
ADEPT: Adaptive Dynamic Ensemble Prediction Techniques |
| Principal Investigator: |
Wyatt, Professor JL |
| Other Investigators: |
|
| Researcher Co-Investigators: |
|
| Project Partners: |
|
| Department: |
School of Computer Science |
| Organisation: |
University of Birmingham |
| Scheme: |
Standard Research |
| Starts: |
01 August 2008 |
Ends: |
31 July 2011 |
Value (£): |
13,017
|
| EPSRC Research Topic Classifications: |
|
| EPSRC Industrial Sector Classifications: |
| No relevance to Underpinning Sectors |
|
|
| Related Grants: |
|
| Panel History: |
| Panel Date | Panel Name | Outcome |
|
04 Sep 2007
|
ICT Prioritisation Panel (Technology)
|
Deferred
|
|
18 Oct 2007
|
ICT Prioritisation Panel (Technology)
|
Announced
|
|
|
Summary on Grant Application Form |
|
Predicting unknown quantities is a fundamental part of science andengineering. For example, in medicine one might wish to predict whether aperson has a cancerous tumour or not based on a scan; or in manufacturing, whether an industrial machine is producing faulty devices or not. The field of Artificial Intelligence has studied many techniques to produce good predictors. The last decade of research has seen the development of population-based techniques. Instead of using a single predictor, these build teams of predictors and combine the decisionsof the individuals through a voting or averaging process. Both theory andexperiments show this reliably improves upon using a single predictor --as they say two heads are better than one. A nice feature is that thesemethods are predictor-independent, meaning they can combine any kind ofpredictors (e.g. neural networks, decisions trees) into a team.This project aims to unify two sub-fields of Artificial Intelligence thatdeal with these population-based predictor-independent techniques:Ensemble Methods and Learning Classifier Systems.Ensemble Methods have produced some of the most powerful predictors ofthe last decade; the most well-known is called AdaBoost , and has beendubbed the best off-the-shelf predictor in the world (Professor LeoBreiman, University of California at Berkeley). These methods have beenwidely applied in many areas; however, one important area not yetinvestigated is multi-step problems. These are problems where decisionsin thepast and present can affect what the best decisions in the future willbe---for example choosing to play a certain opening strategy in chessmeans certain moves are less favourable later on in the game. Our mostdifficult multi-step problem will be optimising elevator scheduling tominimise the amount of time between pressing an elevator call button andthe arrival of the elevator. It is surprisingly difficult to optimise themovement of elevators in a large building. For one thing, a building with5 elevators and 30 floors has more possible configurations than thereare grains of sand on all the beaches in the world. Most ensemble methodscannot be directly applied to this kind of problem.Learning Classifier Systems are a class of nature-inspired algorithms, thatcan dynamically generate and adjust sets of predictors, and are capableof tackling these multi-step problems. Traditional ensemble methodshave not considered the multi-step domain, but have strong theoreticalfoundations to build upon. Learning Classifier Systems do not have sucha strong theory base, but have been intensely studied on multi-step problems.This project will create hybrid methods using theory and practice fromthese two quite disparate fields. We will advance the state-of-the-artin both fields and increase research capacity for tackling several problemclasses, focusing in particular on multi-step problems.
|
|
Key Findings |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
|
Potential use in non-academic contexts |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
|
Impacts |
|
| Description |
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk |
| Summary |
|
| Date Materialised |
|
|
|
|
Sectors submitted by the Researcher |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
| Project URL: |
http://www.cs.man.ac.uk/~gbrown/adept/ |
| Further Information: |
|
| Organisation Website: |
http://www.bham.ac.uk |