Author name cluster
Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.
AAMAS Conference 2023 Conference Paper
ICAPS Conference 2016 Conference Paper
Decentralized POMDP is an expressive model for multi-agent planning. Finite-state controllers (FSCs)---often used to represent policies for infinite-horizon problems---offer a compact, simple-to-execute policy representation. We exploit novel connections between optimizing decentralized FSCs and the dual linear program for MDPs. Consequently, we describe a dual mixed integer linear program (MIP) for optimizing deterministic FSCs. We exploit the Dec-POMDP structure to devise a compact MIP and formulate constraints that result in policies executable in partially-observable decentralized settings. We show analytically that the dual formulation can also be exploited within the expectation maximization (EM) framework to optimize stochastic FSCs. The resulting EM algorithm can be implemented by solving a sequence of linear programs, without requiring expensive message-passing over the Dec-POMDP DBN. We also present an efficient technique for policy improvement based on a weighted entropy measure. Compared with state-of-the-art FSC methods, our approach offers over an order-of-magnitude speedup, while producing similar or better solutions.
AAAI Conference 2013 Conference Paper
The complexity and tight integration of electromechanical systems often makes them “brittle” and hard to modify in response to changing requirements. We aim to remedy this by capturing expert knowledge as functional blueprints, an idea inspired by regulatory processes that occur in natural morphogenesis. We then apply this knowledge in an intelligent design variation tool. When a user modifies a design, our tool uses functional blueprints to modify other components in response, thereby maintaining integration and reducing the need for costly search or constraint solving. In this paper, we refine the functional blueprint concept and discuss practical issues in applying it to electromechanical systems. We then validate our approach with a case study applying our prototype tool to create variants of a miniDroid robot and by empirical evaluation of convergence dynamics of networks of functional blueprints.
AAMAS Conference 2013 Conference Paper
In a distributed information system, Quality of Service (QoS) and Information Assurance (IA) compete for the same set of resources. This tension increases in the presence of cyber attacks. Previous work formulated the problem of trading off QoS against IA as a DCOP whose solution sets the local configuration at individual decision-making nodes to optimize overall levels of QoS and IA delivered by the system. In this paper, we report on the first implementation of maxsum in a realistic distributed system running on multiple machines. Sample results from mission-oriented scenarios based on published documentation and run in an emulated network show the advantage of tradeoff-driven adaptation in meeting QoS and IA requirements.
TIST Journal 2012 Journal Article
We present a novel ensemble architecture for learning problem-solving techniques from a very small number of expert solutions and demonstrate its effectiveness in a complex real-world domain. The key feature of our “Generalized Integrated Learning Architecture” (GILA) is a set of heterogeneous independent learning and reasoning (ILR) components, coordinated by a central meta-reasoning executive (MRE). The ILRs are weakly coupled in the sense that all coordination during learning and performance happens through the MRE. Each ILR learns independently from a small number of expert demonstrations of a complex task. During performance, each ILR proposes partial solutions to subproblems posed by the MRE, which are then selected from and pieced together by the MRE to produce a complete solution. The heterogeneity of the learner-reasoners allows both learning and problem solving to be more effective because their abilities and biases are complementary and synergistic. We describe the application of this novel learning and problem solving architecture to the domain of airspace management, where multiple requests for the use of airspaces need to be deconflicted, reconciled, and managed automatically. Formal evaluations show that our system performs as well as or better than humans after learning from the same training data. Furthermore, GILA outperforms any individual ILR run in isolation, thus demonstrating the power of the ensemble architecture for learning and problem solving.
UAI Conference 2011 Conference Paper
To deal with the prohibitive complexity of calculating policies in Decentralized MDPs, researchers have proposed models that exploit structured agent interactions. Settings where most agent actions are independent except for few actions that affect the transitions and/or rewards of other agents can be modeled using Event-Driven Interactions with Complex Rewards (EDI-CR). Finding the optimal joint policy can be formulated as an optimization problem. However, existing formulations are too verbose and/or lack optimality guarantees. We propose a compact Mixed Integer Linear Program formulation of EDI-CR instances. The key insight is that most action sequences of a group of agents have the same effect on a given agent. This allows us to treat these sequences similarly and use fewer variables. Experiments show that our formulation is more compact and leads to faster solution times and better solutions than existing formulations.
AAMAS Conference 2008 Conference Paper
A database view is a dynamic virtual table composed of the result set of a query, often executed over different underlying databases. The view maintenance problem concerns how a view is refreshed when the data sources are updated. We study the view maintenance problem when self-interested database managers from different institutions are involved, each concerned about the privacy of its database. We regard view maintenance as an incremental, sequential process where an action taken at a stage affects what happens at later stages. The contribution of this paper is twofold. First, we formulate the view maintenance problem as a sequential game of incomplete information where at every stage, each database manager decides what information to disclose, if any, without knowledge of the number or nature of updates at other managers. This allows us to adopt a satisficing approach where the final view need not reflect 100% of the databases updates. Second, we present an anytime algorithm for calculating -Bayes-Nash equilibria that allows us to solve the large games which our problem translates to. Our algorithm is not restricted to games originating from the view maintenance problem; it can be used to solve general games of incomplete information. In addition, experimental results demonstrate our algorithm’s attractive anytime behavior, which allows it to find good-enough solutions to large games within reasonable amounts of time.