Simultaneous Independent Online Auctions with Discrete Bid Increments

Decentralized multi-item auctions offer great opportunities for integrating fragmented online auction markets into larger markets with more efficient outcomes. This paper extends the theory of multi-item ascending auctions of substitutes by considering any finite positive bid increment and allowing the bidders to bid asyn-chronously instead of bidding in a round-robin fashion. We consider a setup where the bidders utilities over multiple items are additive and bound the maximum inefficiency in the allocation when the bidders follow a simple greedy strategy. We also obtain the limits within which the prices of individual items can vary from one outcome to another. For the special case of single unit bidder demand, we also bound the maximum surplus which a bidder can extract by unilaterally switching to some other strategy. The paper suggests an upper bound for the minimum required bid increment which would be necessary for competitive price discovery and truthful bidding in a practical online implementation.The work was done while the author was with IBM Research.     

Probabilistic Automated Bidding in Multiple Auctions

This paper presents an approach to develop bidding agents that participate in multiple auctions with the goal of obtaining an item with a given probability. The approach consists of a prediction method and a planning algorithm. The prediction method exploits the history of past auctions to compute probability functions capturing the belief that a bid of a given price may win a given auction. The planning algorithm computes a price and a set of compatible auctions, such that by sequentially bidding this price in each of the auctions, the agent can obtain the item with the desired probability. Experiments show that the approach increases the payoff of their users and the welfare of the market.     

Knowledge Management in Societies of Intelligent Adaptive Agents

A model is developed of the emergence of the knowledge level in asociety of agents where agents model and manage other agents as resources,and manage the learning of other agents to develop such resources. It isargued that any persistent system that actively creates the conditions forits persistence is appropriately modeled in terms of the rationalteleological models that Newell defines as characterizing the knowledgelevel. The need to distribute tasks in agent societies motivates suchmodeling, and it is shown that if there is a rich order relationship ofdifficulty on tasks that is reasonably independent of agents then it isefficient to model agents competencies in terms of their possessingknowledge. It is shown that a simple training strategy of keeping an agent'sperformance constant by allocating tasks of increasing difficulty as anagent adapts optimizes the rate of learning and linearizes the otherwisesigmoidal learning curves. It is suggested that this provides a basis forassigning a granularity to knowledge that enables learning processes to bemanaged simply and efficiently.     

Adaptive Flocking Control with a Minority of Informed Agents

In this paper, we address the flocking problem of multiple dynamic mobile agents with a virtual leader in a dynamic proximity network. To avoid fragmentation, we propose a novel flocking algorithm that consists of both an adaptive controller for followers and a feedback controller for the virtual leader. Based on our algorithm, all agents in the group can form a network, maintain connectivity, and track the virtual leader, even when only a minority of agents have access to the information of the virtual leader. Finally, several convincing simulation results are provided that demonstrate 2‐D flocking of a group of agents using the proposed algorithm.     

智能机器人的自适应自主Agents的建模

在问题求解期间及穿越动态的、不可预测的环境的过程中 ,真实世界的问题需要自适应的求解方法来‘裁剪’Agent的行为 ,并使其落入任务域 ,否则不完备的知识、不确定性、不确定的Agents及过程的存在、硬件失灵和不精确性都会引发不确定事件的发生。自适应自主Agents是一种存在于动态的、不可预测的环境中 ,并试图满足一系列随时间变化的目标或动机的系统 ,而且此类Agents在处理这些目标的过程中 ,能基于已有的经验改进其能力。自适应自主Agents为智能机器人实现这一目标提供了一种新的途径。主要讨论了智能机器人的自适应自主Agents的建模问题 ,并指出它的局限性和开问题     

Real-Time Search for Autonomous Agents and Multiagent Systems

Since real-time search provides an attractive framework for resource-bounded problem solving, this paper extends the framework for autonomous agents and for a multiagent world. To adaptively control search processes, we propose -search which allows suboptimal solutions with error, and -search which balances the tradeoff between exploration and exploitation. We then consider search in uncertain situations, where the goal may change during the course of the search, and propose a moving target search (MTS) algorithm. We also investigate real-time bidirectional search (RTBS) algorithms, where two problem solvers cooperatively achieve a shared goal. Finally, we introduce a new problem solving paradigm, called organizational problem solving, for multiagent systems.     

Games That Agents Play: A Formal Framework for Dialogues between Autonomous Agents

We present a logic-based formalism for modeling ofdialogues between intelligent and autonomous software agents,building on a theory of abstract dialogue games which we present.The formalism enables representation of complex dialogues assequences of moves in a combination of dialogue games, and allowsdialogues to be embedded inside one another. The formalism iscomputational and its modular nature enables different types ofdialogues to be represented.     

求解组合拍卖NP问题的遗传算法

多智能体系统采用多件物品组合拍卖协商协议能高效地实现组合资源及多任务的分配,但在传统组合拍卖协商协议中拍卖方选择买方以实现最大赢利的过程是一个NP问题,本文提出了利用改进的遗传算法来求解该NP问题的新方法,并应用于一个企业供应链管理的自动协商交易系统中。实验表明,该算法具有较优的性能。     

Auctions in a Two-Sided Network: The Market for Meal Voucher Services

In two-sided networks, two parties interact on a platform, caring about the numbers of subjects on the other side. A typical problem, in this context, is setting prices for network services so as to get “both side on boards”. The standard approach in the literature considers the networks ability to convince both sides to join the network, whereas this paper consider an alternative setting, in which one side determines the price balance. This case is exemplified by the market for meal voucher services in Italy, in which one agency organizes the procurement of services for the public sector, through a competitive auction. A formal model of optimal auction is illustrated and used to assess the current procurement scheme. The model reveals that the current system is inefficient and could be improved through increased flexibility and adaptation to local economic conditions.     

New Algorithms for Mining the Reputation of Participants of Online Auctions

The assessment of credibility and reputation of contractors in online auctions is the key issue in providing reliable environment for customer-to-customer e-commerce. Confident reputation rating system is an important factor in managing risk and building customer satisfaction. Unfortunately, most online auction sites employ a very simple reputation rating scheme that utilizes user feedbacks and comments issued after committed auctions. Such schemes are easy to deceive and do not provide satisfactory protection against several types of fraud. In this paper we propose two novel measures of trustworthiness, namely, credibility and density. We draw inspiration from social network analysis and present two algorithms for reputation rating estimation. Our first algorithm computes the credibility of participants by an iterative search of inter-participant connections. Our second algorithm discovers clusters of participants who are densely connected through committed auctions. We test both measures on real-world data and we experimentally compare them with existing solutions.     

Autonomous Adaptive Control System for Airplane Landing

The paper presents a new autonomous adaptive system for the control of airplanes during landing in longitudinal plane. For the first stage of landing in longitudinal plane, the main variables to be controlled are the glide slope angle and longitudinal velocity; during the second main stage of landing in longitudinal plane, the vertical velocity error and the airplane's longitudinal velocity are controlled. The new robust control architecture has linear subsystems, for which the relative degrees are calculated; the new architecture will also consists of a dynamic compensator, a linear observer, and two reference models, their design being accomplished with respect to the calculated relative degrees. The signal estimated by the observer is useful for training a neural network – an adaptive subsystem of the architecture that provides the adaptive component of the control law. In the case of actuators having nonlinear dynamics, pseudo control hedging blocks are used to cancel the adapting difficulties of the neural networks. The new adaptive architecture is software implemented and validated by complex numerical simulations.     

Inapproximability Results for Combinatorial Auctions with Submodular Utility Functions

We consider the following allocation problem arising in the setting of combinatorial auctions: a set of goods is to be allocated to a set of players so as to maximize the sum of the utilities of the players (i.e., the social welfare). In the case when the utility of each player is a monotone submodular function, we prove that there is no polynomial time approximation algorithm which approximates the maximum social welfare by a factor better than 1−1/e≃0.632, unless P=NP. Our result is based on a reduction from a multi-prover proof system for MAX-3-COLORING. This work was performed when all authors were at the Georgia Institute of Technology. A preliminary version of this work appears in Khot et al. (Workshop on internet and network economics, pp. 92–101, 2005) R.J. Lipton’s research supported by NSF grant CCF-0431023.     

The Advantages of Designing Adaptive Business Agents Using Reputation Modeling Compared to the Approach of Recursive Modeling

Adaptive business agents operate in electronic marketplaces, learning from past experiences to make effective decisions on behalf of their users. How best to design these agents is an open question. In this article, we present an approach for the design of adaptive business agents that uses a combination of reinforcement learning and reputation modeling. In particular, we take into account the fact that multiple selling agents may offer the same good with different qualities, and that selling agents may alter the quality of their goods. We also consider the possibility of dishonest agents in the marketplace. Our buying agents exploit the reputation of selling agents to avoid interaction with the disreputable ones, and therefore to reduce the risk of purchasing low value goods. We then experimentally compare the performance of our agents with those designed using a recursive modeling approach. We are able to show that agents designed according to our algorithms achieve better performance in terms of satisfaction and computational time and as such are well suited for the design of electronic marketplaces.     

C2C网上拍卖中的信任计算模型研究

在线信任匮乏已成为阻碍网上拍卖快速、健康发展的重要因素。为了帮助用户实施有效的信任决策,针对拍卖网站中现有的信任计算方法的不足以及传统理论信任模型的局限性,本文提出了影响网上拍卖信任的5项信任因素,并构建了C2C网上拍卖中的信任计算模型。最后仿真实验的结果表明模型具有较好的有效性和可行性。     

Epistemic Logic for Rule-Based Agents

The logical omniscience problem, whereby standard models of epistemic logic treat an agent as believing all consequences of its beliefs and knowing whatever follows from what else it knows, has received plenty of attention in the literature. But many attempted solutions focus on a fairly narrow specification of the problem: avoiding the closure of belief or knowledge, rather than showing how the proposed logic is of philosophical interest or of use in computer science or artificial intelligence. Sentential epistemic logics, as opposed to traditional possible worlds approaches, do not suffer from the problems of logical omniscience but are often thought to lack interesting epistemic properties. In this paper, I focus on the case of rule-based agents, which play a key role in contemporary AI research but have been neglected in the logical literature. I develop a framework for modelling monotonic, nonmonotonic and introspective rule-based reasoners which have limited cognitive resources and prove that the resulting models have a number of interesting properties. An axiomatization of the resulting logic is given, together with completeness, decidability and complexity results.     

一种基于Agent的自适应的分布式入侵检测系统

针对传统分布式入侵检测系统组件之间依赖程度大、系统不够健壮且入侵检测系统自身结构固定不能适应入侵的变化的问题,提出了一种基于Agent的自适应的分布式入侵检测系统（简称AAADIDS）．AAADIDS采用Agent概念重新构造系统的组件,改进了分布式入侵检测系统由于高层节点单一无冗余而产生的可靠性差的缺陷,从构造上克服了分布式入侵检测系统的脆弱性．同时,AAADIDS系统采用智能技术构建了自适应的入侵检测系统模型,增加了系统应对入侵行为变化的智能性．AAADIDS系统相对于传统的分布式入侵检测系统有效地提高了系统自身的可靠性和针对外界变化的适应能力．     

Software Agents and Their Bodies

Within artificial intelligence and the philosophy of mind,there is considerable disagreement over the relationship between anagent's body and its capacity for intelligent behavior. Some treatthe body as peripheral and tangential to intelligence; others arguethat embodiment and intelligence are inextricably linked. Softwareagents–-computer programs that interact with software environmentssuch as the Internet–-provide an ideal context in which to studythis tension. I develop a computational framework for analyzingembodiment. The framework generalizes the notion of a body beyondmerely having a physical presence. My analysis sheds light oncertain claims made about the relevance of the body to intelligence,as well as on embodiment in software worlds.     

基于自主代理和元胞自动机的交通流模型

提出一种新的元胞自动模型用以模拟单道高速车道的交通流动。一个机动车驾驶单元可以抽象为一个自主代理。模型集中于多种代理的提前相互作用和物理限制(特别是加速、减速能力)。物理限制通过加速减速的双向状态实现。给出了基本图、速度位置图、速度时间图和相位图,并利用互相关函数来辨别自由流和同步流。模拟结果显示这个模型能够描述大部分实际观测的结果。     

一种Internet移动代理系统的黄页服务模型

提出了一种基于信度推理网络的移动代理系统黄页服务模型,为自主迁移的移动代理提供迁移决策支持。与现有的移动代理系统黄页服务模型相比,该模型较好地解决了信息局部性和信息更新的问题,其结构具有良好的可扩展性和灵活性,能更好地适应Ｉｎｔｅｒｎｅｔ动态的网络环境。     

Embedding Connectionist Autonomous Agents in Time: The ‘Road Sign Problem’

This paper identifies a problem of significance for approaches to adaptive autonomous agent research seeking to go beyond reactive behaviour without resorting to hybrid solutions. The feasibility of recurrent neural network solutions are discussed and compared in the light of experiments designed to test ability to handle long-term temporal dependencies, in a more situated context than hitherto. It is concluded that a general-purpose recurrent network with some processing enhancements can begin to fulfil the requirements of this non-trivial problem.