Theoretical Computer Science
Markov Decision Processes: Discrete Stochastic Dynamic Programming
Markov Decision Processes: Discrete Stochastic Dynamic Programming
Approximating the Stochastic Knapsack Problem: The Benefit of Adaptivity
FOCS '04 Proceedings of the 45th Annual IEEE Symposium on Foundations of Computer Science
Stochastic shortest paths via Quasi-convex maximization
ESA'06 Proceedings of the 14th conference on Annual European Symposium - Volume 14
Improved analysis of the greedy algorithm for stochastic matching
Information Processing Letters
Stochastic matching and collaborative filtering to recommend people to people
Proceedings of the fifth ACM conference on Recommender systems
Approximation algorithms for stochastic orienteering
Proceedings of the twenty-third annual ACM-SIAM symposium on Discrete Algorithms
Flow-Based combinatorial chance constraints
CPAIOR'12 Proceedings of the 9th international conference on Integration of AI and OR Techniques in Constraint Programming for Combinatorial Optimization Problems
Stochastic matching with commitment
ICALP'12 Proceedings of the 39th international colloquium conference on Automata, Languages, and Programming - Volume Part I
Catch them if you can: how to serve impatient users
Proceedings of the 4th conference on Innovations in Theoretical Computer Science
A stochastic probing problem with applications
IPCO'13 Proceedings of the 16th international conference on Integer Programming and Combinatorial Optimization
Harnessing the power of two crossmatches
Proceedings of the fourteenth ACM conference on Electronic commerce
User Modeling and User-Adapted Interaction
| Hi-index | 0.00 |
Motivated by applications in online dating and kidney exchange, we study a stochastic matching problem in which we have a random graph G given by a node set V and probabilities p (i ,j ) on all pairs i ,j *** V representing the probability that edge (i ,j ) exists. Additionally, each node has an integer weight t (i ) called its patience parameter. Nodes represent agents in a matching market with dichotomous preferences, i.e., each agent finds every other agent either acceptable or unacceptable and is indifferent between all acceptable agents. The goal is to maximize the welfare, or produce a matching between acceptable agents of maximum size. Preferences must be solicited based on probabilistic information represented by p (i ,j ), and agent i can be asked at most t (i ) questions regarding his or her preferences. A stochastic matching algorithm iteratively probes pairs of nodes i and j with positive patience parameters. With probability p (i ,j ), an edge exists and the nodes are irrevocably matched. With probability 1 *** p (i ,j ), the edge does not exist and the patience parameters of the nodes are decremented. We give a simple greedy strategy for selecting probes which produces a matching whose cardinality is, in expectation, at least a quarter of the size of this optimal algorithm's matching. We additionally show that variants of our algorithm (and our analysis) can handle more complicated constraints, such as a limit on the maximum number of rounds, or the number of pairs probed in each round.