Pragnesh Jay Modi

Adopt: asynchronous distributed constraint optimization with quality guarantees

The Distributed Constraint Optimization Problem (DCOP) is a promising approach for modeling distributed reasoning tasks that arise in multiagent systems. Unfortunately, existing methods for DCOP are not able to provide theoretical guarantees on global solution quality while allowing agents to operate asynchronously. We show how this failure can be remedied by allowing agents to make local decisions based on conservative cost estimates rather than relying on global certainty as previous approaches have done. This novel approach results in a polynomial-space algorithm for DCOP named Adopt that is guaranteed to find the globally optimal solution while allowing agents to execute asynchronously and in parallel. Detailed experimental results show that on benchmark problems Adopt obtains speedups of several orders of magnitude over other approaches. Adopt can also perform bounded-error approximation-it has the ability to quickly find approximate solutions and, unlike heuristic search methods, still maintain a theoretical guarantee on solution quality.

An asynchronous complete method for distributed constraint optimization

We present a new polynomial-space algorithm, called Adopt, for distributed constraint optimization (DCOP). DCOP is able to model a large class of collaboration problems in multi agent systems where a solution within given quality parameters must be found. Existing methods for DCOP are not able to provide theoretical guarantees on global solution quality while operating both efficiently and asynchronously. Adopt is guaranteed to find an optimal solution, or a solution within a user-specified distance from the optimal, while allowing agents to execute asynchronously and in parallel. Adopt obtains these properties via a distributed search algorithm with several novel characteristics including the ability for each agent to make local decisions based on currently available information and without necessarily having global certainty. Theoretical analysis shows that Adopt provides provable quality guarantees, while experimental results show that Adopt is significantly more efficient than synchronous methods. The speedups are shown to be partly due to the novel search strategy employed and partly due to the asynchrony of the algorithm.

A Dynamic Distributed Constraint Satisfaction Approach to Resource Allocation

In distributed resource allocation a set of agents must assign their resources to a set of tasks. This problem arises in many real-world domains such as disaster rescue, hospital scheduling and the domain described in this paper: distributed sensor networks. Despite the variety of approaches proposed for distributed resource allocation, a systematic formalization of the problem and a general solution strategy are missing. This paper takes a step towards this goal by proposing a formalization of distributed resource allocation that represents both dynamic and distributed aspects of the problem and a general solution strategy that uses distributed constraint satisfaction techniques. This paper defines the notion of Dynamic Distributed Constraint Satisfaction Problem (DyDCSP) and proposes two generalized mappings from distributed resource allocation to DyDCSP, each proven to correctly perform resource allocation problems of specific difficulty and this theoretical result is verified in practice by an implementation on a real-world distributed sensor network.

Ariadne: a system for constructing mediators for Internet sources

The Web is based on a browsing paradigm that makes it difficult to retrieve and integrate data from multiple sites. Today, the only way to achieve this integration is by building specialized applications, which are time-consuming to develop and difficult to maintain. We are addressing this problem by creating the technology and tools for rapidly constructing information mediators that extract, query, and integrate data from web sources. The resulting system, called Ariadne, makes it feasible to rapidly build information mediators that access existing web sources.

Dynamic Distributed Resource Allocation: A Distributed Constraint Satisfaction Approach

In distributed resource allocation a set of agents must assign their resources to a set of tasks. This problem arises in many real-world domains such as distributed sensor networks, disaster rescue, hospital scheduling and others. Despite the variety of approaches proposed for distributed resource allocation, a systematic formalization of the problem, explaining the different sources of difficulties, and a formal explanation ofthe strengths and limitations ofk ey approaches is missing. We take a step towards this goal by proposing a formalization of distributed resource allocation that represents both dynamic and distributed aspects ofthe problem. We define four categories ofdif ficulties ofthe problem. To address this formalized problem, the paper defines the notion of Dynamic Distributed Constraint Satisfaction Problem (DyDCSP). The central contribution of the paper is a generalized mapping from distributed resource allocation to DyDCSP. This mapping is proven to correctly perform resource allocation problems of specific difficulty. This theoretical result is verified in practice by an implementation on a real-world distributed sensor network.

Conflicts in teamwork: hybrids to the rescue

Today within the AAMAS community, we see at least four competing approaches to building multiagent systems: belief-desire-intention (BDI), distributed constraint optimization (DCOP), distributed POMDPs, and auctions or game-theoretic approaches. While there is exciting progress within each approach, there is a lack of cross-cutting research. This paper highlights hybrid approaches for multiagent teamwork. In particular, for the past decade, the TEAMCORE research group has focused on building agent teams in complex, dynamic domains. While our early work was inspired by BDI, we will present an overview of recent research that uses DCOPs and distributed POMDPs in building agent teams. While DCOP and distributed POMDP algorithms provide promising results, hybrid approaches help us address problems of scalability and expressiveness. For example, in the BDI-POMDP hybrid approach, BDI team plans are exploited to improve POMDP tractability, and POMDPs improve BDI team plan performance. We present some recent results from applying this approach in a Disaster Rescue simulation domain being developed with help from the Los Angeles Fire Department.

CMRadar: a personal assistant agent for calendar management

Personal assistant agents have long promised to automate routine everyday tasks in order to reduce the cognitive load on humans. One such routine task is the management of a users calendar. In this paper, we describe CMRadar, a calendar management system that is a significant step towards achieving the enduring vision of assistant agents. CMRadar is an implemented system with wide-ranging capabilities for supporting email exchange, multiagent negotiations and schedule optimization based on user preferences. The motivation is to develop an end-to-end system for use by real users to obtain data to facilitate learning. Having now completed an initial prototype which we believe is the first end-to-end agent for calendar management, we present as contributions our architecture design, the communication language used to tie system components together, and initial simulation experiments that isolate negotiation cost a key factor to be logged and predicted in order to improve performance.

Development and Specification of a Reference Model for Agent-Based Systems

Agent-based systems have been the object of intense research over the past decade. While great theoretical progress has been made, the software frameworks for creating agent-based systems offer considerable variability in their capabilities and functionality. This paper introduces a reference model for agent-based systems. The purpose of a reference model is to provide a common conceptual basis for comparing systems and driving the development of software architectures and other standards. The Foundation for Intelligent Physical Agents and other groups have advanced a number of agent standards, yet, to date, no comprehensive reference model has been presented for software systems composed of agents. This paper provides an overview of a reference model for agent-based systems. The agent systems reference model is the result of a multiyear effort studying software systems built with agents and software frameworks for implementing these systems. As part of this study, the team applied software reverse engineering techniques to perform static and dynamic analysis of operational agent-based systems. This analysis enabled identification of key common concepts across over one dozen different agent frameworks. To demonstrate its applicability, the reference model is then used to analyze a number of complete agent-based software systems. It is the belief of the authors that the reference model will be an essential prerequisite for future transition, deployment, and integration of agent-based systems.

Classification of examples by multiple agents with private features

We consider classification tasks where relevant features are distributed among a set of agents and cannot be centralized, for example due to privacy restrictions. We are motivated by a key classification task that arises in a calendar management domain where software assistants classify new meetings as likely to be difficult to schedule. Accurate prediction of the output class is difficult for an isolated single agent because the target concept may involve features to which the agent does not have access, for example each attendees willingness to attend the meeting. To increase prediction accuracy, novel learning algorithms are required in which agents collaborate to classify new examples while maintaining the privacy of features. We introduce a novel distributed asynchronous decision-tree inspired algorithm for such tasks named DDT. DDT differs from previous approaches in that it applies to vertically partitioned data with categorical multi-valued features, it requires no explicit hypothesis generation, and there is no a priori restriction on number of agents. We present empirical results in our meeting scheduling domain and show that DDT outperforms a single agent learner and performs as well as a centralized learner with hypothetical access to all the features.

Collaborative multiagent learning for classification tasks

Multiagent learning differs from standard machine learning in that most existing learning methods assume that all knowledge is available locally in a single agent. In multiagent systems, this assumption does not hold because relevant knowledge is distributed among the agents within the system. We describe a decentralized learning algorithm for {\it distributed classification tasks}, i.e. classification when the attributes are distributed among a set of agents and cannot be gathered into a central agent. Our main contribution is to introduce and formalize the distributed classfication task, show that existing classification algorithms are not satisfactory for distributed classification tasks, and finally, to show that our collaborative learning algorithm performs well at distributed classification.