Zhengzhu Feng

Adaptive peer selection

In a peer-to-peer file-sharing system, a client desiring a particular file must choose a source from which to download. The problem of selecting a good data source is difficult because some peers may not be encountered more than once, and many peers are on low-bandwidth connections. Despite these facts, information obtained about peers just prior to the download can help guide peer selection. A client can gain additional time savings by aborting bad download attempts until an acceptable peer is discovered. We denote as peer selection the entire process of switching among peers and finally settling on one. Our main contribution is to use the methodology of machine learning for the construction of good peer selection strategies from past experience. Decision tree learning is used for rating peers based on low-cost information, and Markov decision processes are used for deriving a policy for switching among peers. Preliminary results with the Gnutella network demonstrate the promise of this approach.

Symbolic Heuristic Search Using Decision Diagrams

We show how to use symbolic model-checking techniques in heuristic search algorithms for both deterministic and decision-theoretic planning problems. A symbolic approach exploits state abstraction by using decision diagrams to compactly represent sets of states and operators on sets of states. In earlier work, symbolic model-checking techniques have been used to find plans that minimize the number of steps needed to reach a goal. Our approach generalizes this by showing how to find plans that minimize the expected cost of reaching a goal. For this generalization, we use algebraic decision diagrams instead of binary decision diagrams. In particular, we show that algebraic decision diagrams provide a compact representation of state evaluation functions. We describe symbolic generalizations of A* search for deterministic planning and of LAO* search for decision-theoretic planning problems formalized as Markov decision processes.We report experimental results and discuss issues for future work.