In this paper we discuss a number of previously unaddressed issues that arise in automated negotiation among self-interested agents whose rationality is bounded by computational complexity. These issues are presented in the context of iterative task allocation negotiations. First, the reasons why such agents need to be able to choose the stage and level of commitment dynamically are identified. A protocol that allows such choices through conditional commitment breaking penalties is presented. Next, the implications of bounded rationality are analyzed. Several tradeoffs between allocated computation and negotiation benefits and risk are enumerated, and the necessity of explicit local deliberation control is substantiated. Techniques for linking negotiation items and multiagent contracts are presented as methods for escaping local optima in the task allocation process. Implementing both methods among self-interested bounded rational agents is discussed. Finally, the ...

This paper analyzes coalitions among self-interested agents that need to solve combinatorial optimization problems to operate e ciently in the world. By colluding (coordinating their actions by solving a joint optimization prob-lem) the agents can sometimes save costs compared to operating individually. A model of bounded rationality is adopted where computation resources are costly. It is not worthwhile solving the problems optimally: solution quality is decision-theoretically traded o against computation cost. A normative, application- and protocol-independent theory of coalitions among bounded-rational agents is devised. The optimal coalition structure and its stability are signi cantly a ected by the agents ' algorithms ' performance pro les and the cost of computation. This relationship is rst analyzed theoretically. Then a domain classi cation including rational and bounded-rational agents is in-troduced. Experimental results are presented in vehicle routing with real data from ve dispatch centers. This problem is NP-complete and the instances are so large that|with current technology|any agent's rationality is bounded by computational complexity. 1

A Dissertation Presented by TUOMAS W. SANDHOLM

We analyze task reallocation where individually rational (IR) agents (re)contract tasks among themselves based on marginal costs. A task allocation graph is introduced as a tool for analyzing contract types. Traditional single task contracts always have a short path (sequence of contracts) to the optimal task allocation but an IR path may not exist, or it may not be short. We analyze an algorithm for finding the shortest IR path. Next we introduce cluster contracts, swaps, and multiagent contracts. Each of the four contract types avoids some local optima that the others do not. Even if the protocol is equipped with all four types, local optima exist. To attack this problem, we introduce OCSMcontracts which combine the ideas behind the four earlier types into an atomic contract type. If the protocol is equipped with OCSM-contracts, any sequence of IR contracts leads to the optimal task allocation in a finite number of steps: an oracle---or speculation---is not needed for choosing the pa...

This paper analyzes coalition formation among self-interested agents that need to solve combinatorial optimization problems to operate efficiently in the world. By colluding (coordinating their actions by solving a joint optimization problem), the agents can sometimes save costs compared to operating individually. A model of bounded rationality is adopted, where computation resources are costly. It is not worth solving the problems optimally: solution quality is decision-theoretically traded off against computation cost. A normative theory of coalitions among bounded rational (BR) agents is devised. The optimal coalition structure and its stability are significantly affected by the agents ' algorithms ' performance profiles (PPs) and the cost of computation. This relationship is first analyzed theoretically. A domain classification including rational and BR agents is introduced. Experimental results are presented in the distributed vehicle routing domain using real data from 5 dispatch centers; the optimal coalition structure for BR agents differs significantly from the one for rational agents. These problems are NP-complete and the instances are so large that, with current technology, any agent's rationality is bounded by computational complexity. 1

Abstract. The representations currently used by local search and some evoluationary algorithms have the disadvantage that these algorithms are partially blind to “ridges ” in the search space. Both heuristics search and gradient search algorithms can exhibit extremely slow convergence on functions that display ridge structures. A class of rotated representations are proposed; these rotated representations can be based on Principal Components Analysis, or use the Gram-Schmidt orthogonalization method. Preliminary experiments show that local search using a rotated representation is able to align the coordinates of the search with ridge structures. Search using a rotated representation can convergence are as much as 1000 times faster than local search using a fixed coordinate representation. 1

Several factors contribute to making search problems easy or difficult. One of these factors is the modality of the fitness landscape. Quite often, when local search algorithms fail to locate the global optimum, it is because the algorithm converged to a local optimum. The majority of local search methods in use today maneuver through the search space using local neighborhood information around a single point to guide the search. In that search paradigm, the number of local optima that occur in the search space has a tremendous effect on search performance. Genetic

This dissertation is dedicated to my mom, Maria, whose love and support made it possible. ACKNOWLEDGMENTS Eliot Moss has been a great thesis advisor. He has helped me to become a better re-searcher by shaping my critical thinking as well as by improving my expressive skills. I would like to thank the members of my thesis committee, Andy Barto, Emery Berger, and Wayne Burleson for their feedback and advice that helped to improve the overall quality of this dissertation. I gratefully acknowledge the friendships and interactions from all members of the Ar-chitecture and Language Implementation group (ALI). Beginning with my first lab meeting talk, I have received helpful feedback on the best way to present myself and my work. The ongoing discussions in the lab helped to stimulate my research. Thanks especially to M. Tyler Maxwell for some of the amazing diagrams in this dissertation. Robbie Moll was helpful at stimulating my research interests in the applications of machine learning and for believing in me as an instructor. I especially would like to acknowledge Emmanuel Agu, who has been a good friend and with whom I have had many rewarding discussions on research and life. Finally, I am extremely grateful for the love and support of my entire family. Overall, I am extremely lucky to be part of such a close and wonderful family. I would like to express my sincerest gratitude to my mother, Maria. As a young child I remember my mother always telling me that I could accomplish anything that I set my mind to. She was right as always. Her confidence in me gave me the strength both to overcome any difficulties and to maintain high goals. This work was supported by National Physical Science Consortium and Lawrence Liv-ermore National Laboratory.

. It is a broadly accepted fact that evolutionary algorithms (EA) have to be developed problem--specifically. Usually this is based on experience and experiments. Though, most EA environments are not suited for such an approach. Therefore, this paper proposes a few basic concepts which should be supplied by modern EA simulators in order to serve as a toolkit for the development of such algorithms. 1 Introduction Theoretical work as well as practical experience demonstrate the importance to progress from fixed, rigid schemes of evolutionary algorithms (EA) towards a problem--specific processing of optimization problems. Since the "No Free Lunch" theorem [WM97] proves that there is no algorithm which performs better than any other algorithm for all kinds of possible problems, it is useless to judge an algorithm irrespectively of the optimization problem. Therefore, it is necessary to find a suitable algorithm for each problem. Experience has shown that the adaptation of a problem ...

By analyzing mutation operators' behavior through their effect on the tness only, fundamental statements on the interplay of the operator's quality and the state of the search process are enabled for arbitrary optimization problems with arbitrary discrete search spaces. First, it is observed that the defined operator quality has a direct effect on hill climbing search. Moreover, an operator's locality property influences its quality depending on the fitness level. This leads to the major result that a local operator cannot be valued without considering the search state.