ESDEMUU
Efficient sequential decision making under uncertainty
| Coordinatore | ECOLE POLYTECHNIQUE FEDERALE DE LAUSANNE
Organization address
address: BATIMENT CE 3316 STATION 1 contact info |
| Nazionalità Coordinatore | Switzerland [CH] |
| Totale costo | 232˙777 € |
| EC contributo | 232˙777 € |
| Programma | FP7-PEOPLE
Specific programme "People" implementing the Seventh Framework Programme of the European Community for research, technological development and demonstration activities (2007 to 2013) |
| Code Call | FP7-PEOPLE-2010-IEF |
| Funding Scheme | MC-IEF |
| Anno di inizio | 2011 |
| Periodo (anno-mese-giorno) | 2011-05-01 - 2013-04-30 |
Partecipanti
| # | ||||
|---|---|---|---|---|
| 1 |
ECOLE POLYTECHNIQUE FEDERALE DE LAUSANNE
Organization address
address: BATIMENT CE 3316 STATION 1 contact info |
CH (LAUSANNE) | coordinator | 232˙777.80 |
Mappa
Word cloud
Esplora la "nuvola delle parole (Word Cloud) per avere un'idea di massima del progetto.
Obiettivo del progetto (Objective)
'Many applications require efficient methods for automated decision making, such as control systems, crisis response, finance, logistics, network security, robotics and traffic management. These problems involve sequential learning and decision making under uncertainty in an unknown environment. As we have incomplete information about the state and dynamics of the environment, the outcome of any specific plan is uncertain. Statistical decision theory offers a framework for finding optimal solutions, but in most problems of interest exact inference and planning are intractable.
The project will develop efficient approximate methods for nearly optimal learning and decision making in such problems. Our first goal is to obtain provably efficient algorithms for decision making in discrete, fully observable environments. Our second goal is to extend these to continuous and partially observable domains. Recent advances in statistical learning theory and in stochastic planning, make this avenue of research particularly promising. Our third theoretical goal is to consider collaborative planning among multiple agents in unknown environments for each of the above cases.
Finally, we shall develop open source code and perform extensive comparative experiments in classical benchmark problems for evaluation purposes. As a more realistic test-bed, we shall focus on the network intrusion detection and response problem, where we must safeguard a network against the attacks of malicious users.
The project coordinator is an expert on Bayesian reinforcement learning and stochastic planning and the host institution has produced seminal breakthroughs in the area of distributed planning, while both have prior experience in problems of network intrusion detection.'