Results 1  10
of
19
Distributed constraint satisfaction and optimization with privacy enforcement
 In 3rd IC on Intelligent Agent Technology
, 2004
"... Several naturally distributed negotiation/cooperation problems with privacy requirements can be modeled within the distributed constraint satisfaction framework, where the constraints are secrets of the participants. Most of the existing techniques aim at various tradeoffs between complexity and pri ..."
Abstract

Cited by 49 (10 self)
 Add to MetaCart
(Show Context)
Several naturally distributed negotiation/cooperation problems with privacy requirements can be modeled within the distributed constraint satisfaction framework, where the constraints are secrets of the participants. Most of the existing techniques aim at various tradeoffs between complexity and privacy guarantees, while others aim to maximize privacy first [12, 7, 3, 4, 11]. In [7] we introduced a first technique allowing agents to solve distributed constraint problems (DisCSPs), without revealing anything and without trusting each other or some server. The technique we propose now is a dm times improvement for m variables of domain size d. On the negative side, the fastest versions of the new technique require storing of O(d m) big integers. From a practical point of view, we improve the privacy with which these problems can be solved, and improve the efficiency with which ⌊n−1/2⌋privacy can be achieved, while it remains inapplicable for larger problems. The technique of [7] has a simple extension to optimization for distributed weighted CSPs. However, that obvious extension leaks to everybody sensitive information concerning the quality of the computed solution. We found a way to avoid this leak, which constitutes another contribution of this paper. 1.
Constraintbased reasoning and privacy/efficiency tradeoffs in multiagent problem solving
 Artificial. Intelligence
, 2005
"... problem solving ..."
(Show Context)
Meeting scheduling guaranteeing n/2privacy and resistant to statistical analysis (applicable to any discsp
 In 3rd IC on Web Intelligence
, 2004
"... Distributed problems raise privacy issues. The user would like to specify securely his constraints (desires, availability, money) on his computer once. The computer is expected to compute and communicate for searching an acceptable solution while maintaining the privacy of the user. Even without com ..."
Abstract

Cited by 13 (2 self)
 Add to MetaCart
(Show Context)
Distributed problems raise privacy issues. The user would like to specify securely his constraints (desires, availability, money) on his computer once. The computer is expected to compute and communicate for searching an acceptable solution while maintaining the privacy of the user. Even without computers infested with spy viruses that capture the interaction with the user, most agent based approaches reveal parts of one agent’s secret data to its partners in distributed computations [7]. Some cryptographic multiparty computation protocols [1] succeed to avoid leaking secrets at the computation of some functions with private inputs. They have been applied to find the set of all solutions for the meeting scheduling problem [3]. However, nobody yet succeeded to apply those techniques for finding a random solution to the meeting scheduling problem. Note that revealing all solutions, when you only need a single one, leaks a lot of data about when others are, or are not, available. Some answers were proposed in our previous approaches to distributed constraint problems [4]. They guarantee that no agent can infer with certitude a secret from the identity of the solution of the problem (other than the acceptance of the solution), but guarantee nothing about inference of probabilistic information about secrets. Our new technique answers this problem, too. 1.
The effect of policies for selecting the solution of a DisCSP on privacy loss
 In AAMAS
, 2004
"... ..."
Incentive auctions and stable marriages problems solved with n/2privacy of human preferences
, 2004
"... Incentive auctions let several participants to cooperate for clearing a set of offers and requests, ensuring that each participant cannot do better than by inputing his true utility. This increases the social welfare by efficient allocations, and is proven to have similar outcomes as the traditional ..."
Abstract

Cited by 10 (8 self)
 Add to MetaCart
Incentive auctions let several participants to cooperate for clearing a set of offers and requests, ensuring that each participant cannot do better than by inputing his true utility. This increases the social welfare by efficient allocations, and is proven to have similar outcomes as the traditional English Auctions. The deskmates (stable matchings) problem comes from the need of placing students in pairs of two for working in projects or seating in twoseats desks. The stable marriages problem consists of finding matches of a man and a woman out of two sets of men, respectively women. Each of the persons in the previous two problems has a (hopefully stable) secret preference between every two possible partners. The participants want to find an allocation satisfying their secret preferences and without leaking any of these secret preferences, except for what a participant can infer from the identity of the partner/spouse that was recommended to her/him. We use a distributed weighted constraint satisfaction (DisWCSP) framework where the actual constraints are secrets that are not known by any agent. They are defined by a set of functions on some secret inputs from all agents. The solution is also kept secret and each agent learns just the result of applying an agreed function on the solution. The new framework is shown to improve the efficiency in modeling the aforementioned problems. We show how to extend our previous techniques to solve securely problems modeled with the new formalism, and exemplify with the two problems in the title. 1 1
Deskmates (stable matching) with privacy of preferences, and a new distributed CSP framework
 In Proc. of CP’2004 Immediate Applications of Constraint Programming Workshop
"... The deskmates matcher application places students in pairs of two for working in projects (similar to the well known problems of stable matchings or stable roommates). Each of the students has a (hopefully stable) secret preference between every two colleagues. The participants want to find an allo ..."
Abstract

Cited by 9 (4 self)
 Add to MetaCart
(Show Context)
The deskmates matcher application places students in pairs of two for working in projects (similar to the well known problems of stable matchings or stable roommates). Each of the students has a (hopefully stable) secret preference between every two colleagues. The participants want to find an allocation satisfying their secret preferences and without leaking any of these secret preferences, except for what a participant can infer from the identity of the partner that was recommended to her. The peculiarities of the above problem require solvers based on old distributed CSP frameworks to use models whose search spaces are larger than those in centralized solvers, with bad effects on efficiency. Therefore we introduce a new distributed constraint satisfaction (DisCSP) framework where the actual constraints are secrets that are not known by any agent. They are defined by a set of functions on some secret inputs from all agents. The solution is also kept secret and each agent learns just the result of applying an agreed function on the solution. The expressiveness of the new framework is shown to improve the efficiency (O(2 m3−log(m)) times) in modeling and solving the aforementioned problem with m participants. We show how to extend our previous techniques to solve securely problems modeled with the new formalism, and exemplify with the problem in the title. An experimental implementation in the form of an appletbased solver is available.
ADOPTing: Unifying Asynchronous Distributed Optimization with Asynchronous Backtracking
, 2007
"... This article presents an asynchronous algorithm for solving Distributed Constraint Optimization problems (DCOPs). The proposed technique unifies asynchronous backtracking (ABT) and asynchronous distributed optimization (ADOPT) where valued nogoods enable more flexible reasoning and more opportunitie ..."
Abstract

Cited by 4 (1 self)
 Add to MetaCart
(Show Context)
This article presents an asynchronous algorithm for solving Distributed Constraint Optimization problems (DCOPs). The proposed technique unifies asynchronous backtracking (ABT) and asynchronous distributed optimization (ADOPT) where valued nogoods enable more flexible reasoning and more opportunities for communication, leading to an important speedup. While feedback can be sent in ADOPT by COST messages only to one predefined predecessor, our extension allows for sending such information to any relevant agent. The concept of valued nogood is an extension by Dago and Verfaille of the concept of classic nogood that associates the list of conflicting assignments with a cost and, optionally, with a set of references to culprit constraints. DCOPs have been shown to have very elegant distributed solutions, such as ADOPT, distributed asynchronous overlay (DisAO), or DPOP. These algorithms are typically tuned to minimize the longest causal chain of messages as a measure of how the algorithms will scale for systems with remote agents (with large latency in communication). ADOPT has the property of maintaining the initial distribution of the problem. To be efficient, ADOPT needs a preprocessing step consisting of computing a DepthFirst Search (DFS) tree on the constraint graph. Valued nogoods allow for automatically detecting and exploiting the best DFS tree compatible with the current ordering. To exploit such DFS trees it is now sufficient to ensure that they exist. Also, the inference rules available for valued nogoods help to exploit schemes of communication where more feedback is sent to higher priority agents. Together they result in an order of magnitude improvement.
Asynchronous aggregation and consistency in distributed . . .
, 2004
"... Constraint Satisfaction Problems (CSP) have been very successful in problemsolving tasks ranging from resource allocation and scheduling to configuration and design. Increasingly, many of these tasks pose themselves in a distributed setting where variables and constraints are distributed among diff ..."
Abstract

Cited by 1 (0 self)
 Add to MetaCart
(Show Context)
Constraint Satisfaction Problems (CSP) have been very successful in problemsolving tasks ranging from resource allocation and scheduling to configuration and design. Increasingly, many of these tasks pose themselves in a distributed setting where variables and constraints are distributed among different agents. A variety of asynchronous search algorithms have been proposed for addressing this setting. We show how two techniques commonly used in centralized constraint satisfaction, value aggregation and maintaining arc consistency can be applied to increase efficiency in an asynchronous, distributed context as well, and report on experiments that quantify the gains.
Discussion on the Three Backjumping Schemes Existing in ADOPTng
"... Abstract. The original ADOPTng has three major versions, corresponding to three different classes of feedback possibilities. The first version is identical to the scheme of the original ADOPT, where messages with feedback are communicated only to the variable of one’s parent node in the DFS of the ..."
Abstract

Cited by 1 (0 self)
 Add to MetaCart
(Show Context)
Abstract. The original ADOPTng has three major versions, corresponding to three different classes of feedback possibilities. The first version is identical to the scheme of the original ADOPT, where messages with feedback are communicated only to the variable of one’s parent node in the DFS of the constraint graph. It is similar to the GraphBased Backjumping concept common in Constraint Satisfaction (CSPs), except that the asynchronous computation paradigm makes the term “backjumping ” less intuitively accurate. The second major version of ADOPTng communicates costs to higher priority agents based on dependencies detected dynamically. The third version combined dependencies detected dynamically with statically analyzed constraint graph structure. These versions are related to ConflictBased Backjumping schemes in CSPs in the way conflicts are announced to earlier variables. Here we discuss and experiment in more detail the advantages and drawbacks of the different backjumping schemes and of some of their variations. While past experiments have shown that sending more feedback is better than sending the minimal information needed for correctness, new experiments show that one should not exaggerate sending too much feedback and that the best strategy is at an intermediary point. 1
NITECH,
"... We merge two popular optimization criteria of Distributed Constraint Optimization Problems (DCOPs) – rewardbased utility and privacy – into a single criterion. Privacy requirements on constraints has classically motivated an optimization criterion of minimizing the number of disclosed tuples, or ma ..."
Abstract
 Add to MetaCart
(Show Context)
We merge two popular optimization criteria of Distributed Constraint Optimization Problems (DCOPs) – rewardbased utility and privacy – into a single criterion. Privacy requirements on constraints has classically motivated an optimization criterion of minimizing the number of disclosed tuples, or maximizing the entropy about constraints. Common complete DCOP search techniques seek solutions minimizing the cost and maintaining some privacy. We start from the observation that for some problems we could provide as input a quantification of loss of privacy in terms of cost. We provide a formal way to integrate this new input parameter into the DCOP framework, discuss its implications and advantages. 1