TY - GEN
T1 - Letting loose a SPIDER on a network of POMDPs
T2 - 6th International Joint Conference on Autonomous Agents and Multiagent Systems, AAMAS'07
AU - Varakantham, Pradeep
AU - Marecki, Janusz
AU - Yabu, Yuichi
AU - Tambe, Milind
AU - Yokoo, Makoto
PY - 2007/12/1
Y1 - 2007/12/1
N2 - Distributed Partially Observable Markov Decision Problems (Distributed POMDPs) are a popular approach for modeling multi-agent systems acting in uncertain domains. Given the significant complexity of solving distributed POMDPs, particularly as we scale up the numbers of agents, one popular approach has focused on approximate solutions. Though this approach is efficient, the algorithms within this approach do not provide any guarantees on solution quality. A second less popular approach focuses on global optimality, but typical results are available only for two agents, and also at considerable computational cost. This paper overcomes the limitations of both these approaches by providing SPIDER, a novel combination of three key features for policy generation in distributed POMDPs: (i) it exploits agent interaction structure given a network of agents (i.e. allowing easier scale-up to larger number of agents); (ii) it uses a combination of heuristics to speedup policy search; and (iii) it allows quality guaranteed approximations, allowing a systematic tradeoff of solution quality for time. Experimental results show orders of magnitude improvement in performance when compared with previous global optimal algorithms.
AB - Distributed Partially Observable Markov Decision Problems (Distributed POMDPs) are a popular approach for modeling multi-agent systems acting in uncertain domains. Given the significant complexity of solving distributed POMDPs, particularly as we scale up the numbers of agents, one popular approach has focused on approximate solutions. Though this approach is efficient, the algorithms within this approach do not provide any guarantees on solution quality. A second less popular approach focuses on global optimality, but typical results are available only for two agents, and also at considerable computational cost. This paper overcomes the limitations of both these approaches by providing SPIDER, a novel combination of three key features for policy generation in distributed POMDPs: (i) it exploits agent interaction structure given a network of agents (i.e. allowing easier scale-up to larger number of agents); (ii) it uses a combination of heuristics to speedup policy search; and (iii) it allows quality guaranteed approximations, allowing a systematic tradeoff of solution quality for time. Experimental results show orders of magnitude improvement in performance when compared with previous global optimal algorithms.
UR - http://www.scopus.com/inward/record.url?scp=60349101997&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=60349101997&partnerID=8YFLogxK
U2 - 10.1145/1329125.1329388
DO - 10.1145/1329125.1329388
M3 - Conference contribution
AN - SCOPUS:60349101997
SN - 9788190426275
T3 - Proceedings of the International Conference on Autonomous Agents
SP - 822
EP - 829
BT - AAMAS'07 - Proceedings of the 6th International Joint Conference on Autonomous Agents and Multiagent Systems
Y2 - 14 May 2008 through 18 May 2008
ER -