Results 1 - 10
of
12
Practical voting rules with partial information
- AUTON AGENT MULTI-AGENT SYST
, 2010
"... Voting is an essential mechanism that allows multiple agents to reach a joint decision. The joint decision, representing a function over the preferences of all agents, is the winner among all possible (candidate) decisions. To compute the winning candidate, previous work has typically assumed that ..."
Abstract
-
Cited by 21 (4 self)
- Add to MetaCart
Voting is an essential mechanism that allows multiple agents to reach a joint decision. The joint decision, representing a function over the preferences of all agents, is the winner among all possible (candidate) decisions. To compute the winning candidate, previous work has typically assumed that voters send their complete set of preferences for computation, and in fact this has been shown to be required in the worst case. However, in practice, it may be infeasible for all agents to send a complete set of preferences due to communication limitations and willingness to keep as much information private as possible. The goal of this paper is to empirically evaluate algorithms to reduce communication on various sets of experiments. Accordingly, we propose an iterative algorithm that allows the agents to send only part of their preferences, incrementally. Experiments with simulated and real-world data show that this algorithm results in an average of 35 % savings in communications, while guaranteeing that the actual winning candidate is revealed. A second algorithm applies a greedy heuristic to save up to 90 % of communications. While this heuristic algorithm cannot guarantee that a true winning candidate is found, we show that in practice, close approximations are obtained.
A Computational Decision Theory for Interactive Assistants
"... We study several classes of interactive assistants from the points of view of decision theory and computational complexity. We first introduce a special class of POMDPs called hidden-goal MDPs (HGMDPs), which formalize the problem of interactively assisting an agent whose goal is hidden and whose ac ..."
Abstract
-
Cited by 8 (0 self)
- Add to MetaCart
(Show Context)
We study several classes of interactive assistants from the points of view of decision theory and computational complexity. We first introduce a special class of POMDPs called hidden-goal MDPs (HGMDPs), which formalize the problem of interactively assisting an agent whose goal is hidden and whose actions are observable. In spite of its restricted nature, we show that optimal action selection in finite horizon HGMDPs is PSPACEcomplete even in domains with deterministic dynamics. We then introduce a more restricted model called helper action MDPs (HAMDPs), where the assistant’s action is accepted by the agent when it is helpful, and can be easily ignored by the agent otherwise. We show classes of HAMDPs that are complete for PSPACE and NP along with a polynomial time class. Furthermore, we show that for general HAMDPs a simple myopic policy achieves a regret, compared to an omniscient assistant, that is bounded by the entropy of the initial goal distribution. A variation of this policy is also shown to achieve worst-case regret that is logarithmic in the number of goals for any goal distribution.
Applying MDP approaches for estimating outcome of interaction in collaborative human-computer settings
- In MSDM 2007
, 2007
"... This paper investigates the problem of determining when a computer agent should interrupt a person with whom it is working collaboratively as part of a distributed, multi-agent team, which is operating in environments in which conditions may be rapidly changing, actions occur at a fast pace, and dec ..."
Abstract
-
Cited by 6 (4 self)
- Add to MetaCart
(Show Context)
This paper investigates the problem of determining when a computer agent should interrupt a person with whom it is working collaboratively as part of a distributed, multi-agent team, which is operating in environments in which conditions may be rapidly changing, actions occur at a fast pace, and decisions must be made within tightly constrained time frames. An interruption would enable the agent to obtain information useful for performing its role in the team task, but the person will incur a cost in responding. The paper presents a formalization of interruptions as multi-agent decision making. It defines a novel, efficient approximation method that decouples the multi-agent decision model into separate MDPs, thereby overcoming the complexity of finding optimal solutions of the Dec-POMDP model. For singleshot situations, the separate outcomes can be combined to give an exact value for the interruption. In more general settings, the closeness of the approximation to the optimal solution depends on the structure of the problem. The paper describes domain specific heuristic functions that improve the efficiency of the approximation further for a specific application. 1.
Collaborative health care plan support
- In Proceedings of the 12th international conference on Autonomous agents and multi-agent systems
, 2013
"... (Article begins on next page) The Harvard community has made this article openly available. Please share how this access benefits you. Your story matters. ..."
Abstract
-
Cited by 6 (3 self)
- Add to MetaCart
(Show Context)
(Article begins on next page) The Harvard community has made this article openly available. Please share how this access benefits you. Your story matters.
An analysis of privacy loss in k-optimal algorithms
- In DCR
, 2008
"... Abstract.For agents to be trusted with sensitive data, they must have mechanisms to protect their users ’ privacy. This has been recognized and addressed in several multi-agent algorithms for distributed constraint optimization. However, privacy concerns are often even more salient in situations whe ..."
Abstract
-
Cited by 3 (0 self)
- Add to MetaCart
(Show Context)
Abstract.For agents to be trusted with sensitive data, they must have mechanisms to protect their users ’ privacy. This has been recognized and addressed in several multi-agent algorithms for distributed constraint optimization. However, privacy concerns are often even more salient in situations where the problem is so large and/or dynamic that complete algorithms are infeasible. This paper explores the privacy properties of k-optimal algorithms: those algorithms that produce locally optimal solutions that can not be improved by changing the assignments of k or fewer agents. This paper shows that while these algorithms are subject to large amounts of privacy loss as presented–particularly as k increases–they can be modified to reduce this privacy loss considerably. The greatest improvements are achieved by replacing the centralized local search with a distributed algorithm, such as DPOP. Using these methods can reduce the privacy loss experienced by agents in these algorithms by an order of magnitude, rendering them appropriate for use in settings where privacy matters. 1
CRISP -- An Interruption Management Algorithm based on Collaborative Filtering
- CHI
, 2014
"... Interruptions can have a significant impact on users working to complete a task. When people are collaborating, either with other users or with systems, coordinating interruptions is an important factor in maintaining efficiency and preventing information overload. Computer systems can observe user ..."
Abstract
-
Cited by 3 (3 self)
- Add to MetaCart
(Show Context)
Interruptions can have a significant impact on users working to complete a task. When people are collaborating, either with other users or with systems, coordinating interruptions is an important factor in maintaining efficiency and preventing information overload. Computer systems can observe user behavior, model it, and use this to optimize the interruptions to minimize disruption. However, current techniques often re-quire long training periods that make them unsuitable for on-line collaborative environments where new users frequently participate. In this paper, we present a novel synthesis between Collaborative Filtering methods and machine learning classification algorithms to create a fast learning algorithm, CRISP. CRISP exploits the similarities between users in order to apply data from known users to new users, therefore requiring less in-formation on each person. Results from user studies indicate the algorithm significantly improves users ’ performances in completing the task and their perception of how long it took to complete each task.
Leveraging Users for Efficient Interruption Management in Agent–User Systems
, 2009
"... In collaborative systems involving a user and an agent working together on a joint task it may be important to share information in order to determine the appropriate course of action. However, communication between agents and users can create costly user interruptions. One of the most important iss ..."
Abstract
-
Cited by 1 (1 self)
- Add to MetaCart
In collaborative systems involving a user and an agent working together on a joint task it may be important to share information in order to determine the appropriate course of action. However, communication between agents and users can create costly user interruptions. One of the most important issue concerning the initiation of information sharing in collaborative systems is the ability to accurately estimate the cost and benefit arising from those interruptions. While cost estimation of interruptions has been previously investigated, these works assumed either a large amount of information was available about each user, or only a small number of states needed consideration. This paper presents a novel synthesis between Collaborative Filtering methods with classi cation algorithms tools to create a fast learning algorithm, MICU. MICU exploits the similarities between users in order to learn from known users to new but similar users and therefore requires less information on each user in compare to other methods. Experimental results indicate the algorithm significantly improves system performance even with a small amount of data on each user.
Iterative Voting Rules
, 2010
"... Voting is an essential mechanism that allows multiple agents to reach a joint decision. The joint decision, representing a function over the preferences of all agents, is the winner among all possible (candidate) decisions. To compute the winning candidate, previous work has typically assumed that v ..."
Abstract
- Add to MetaCart
Voting is an essential mechanism that allows multiple agents to reach a joint decision. The joint decision, representing a function over the preferences of all agents, is the winner among all possible (candidate) decisions. To compute the winning candidate, previous work has typically assumed that voters send their complete set of preferences for computation, and in fact this has been shown to be required in the worst case. However, in practice, it may be infeasible for all agents to send a complete set of preferences due to communication limitations and willingness to keep as much information private as possible. The goal of this paper is to empirically evaluate algorithms to reduce communication on various sets of experiments. Accordingly, we propose an iterative algorithm that allows the agents to send only part of their preferences, incrementally. Experiments with simulated and real-world data show that this algorithm results in an average of 35 % savings in communications, while guaranteeing that the actual winning candidate is revealed. A second algorithm applies a greedy heuristic to save up to 90 % of communications. While this heuristic algorithm cannot guarantee that a true winning candidate is found, we show that in practice, close approximations are obtained.
UNIVERSITY OF SOUTHAMPTON
"... Bayesian learning for multi-agent coordination by ..."
(Show Context)
