Catálogo de publicaciones - libros

The “ZIP” adaptive trading algorithm has been demonstrated to outperform human traders in experimental studies of continuous double auction (CDA) markets. The original ZIP algorithm requires the values of eight control parameters to be set correctly. A new extension of the ZIP algorithm, called ZIP60, requires the values of 60 parameters to be set correctly. ZIP60 is shown here to produce significantly better results than the original ZIP (called “ZIP8” hereafter), for negligable additional computational costs. A genetic algorithm (GA) is used to search the 60-dimensional ZIP60 parameter space, and it finds parameter vectors that yield ZIP60 traders with mean scores significantly better than those of ZIP8s. This paper shows that the optimizing evolutionary search works best when the GA itself controls the dimensionality of the search-space, so that the search commences in an 8-d space and thereafter the dimensionality of the search-space is gradually increased by the GA until it is exploring a 60-d space. Furthermore, the results from ZIP60 cast some doubt on prior ZIP8 results concerning the evolution of new ‘hybrid’ auction mechanisms that appeared to be better than the CDA.

In a previous work we extended the notion of multi-unit combinatorial reverse auction (MUCRA) by adding a new dimension to the goods at auction. A buyer can express transformability relationships among goods: some goods can be transformed into others at a transformation cost. Through this new auction type, a buyer can find out what goods to buy, to whom, and what transformations to apply to the acquired goods in order to obtain the best savings. The main focus of the paper is to perform some preliminary experiments to quantitatively assess the potential savings that a buying agent may obtain in considering transformation relationships.

This paper studies bilateral, multi-issue negotiation between self interested agents with deadlines. There are a number of procedures for negotiating the issues and each of these gives a different outcome. Thus, a key problem is to decide which one to use. Given this, we study the three main alternatives: the , the , and the . First, we determine equilibria for the case where each agent is uncertain about its opponent’s deadline. We then compare the outcomes for these procedures and determine the one that is optimal (in this case, the package deal is optimal for each party). We then compare the procedures in terms of their time complexity, the uniqueness and Pareto optimality of their solutions, and their time of agreement.

Supply chains are ubiquitous in the manufacturing of many complex products. Traditionally, supply chains have been created through the interactions of human representatives of the companies involved, but advances in autonomous agent technologies have sparked an interest in automating the process. The Trading Agent Competition Supply Chain Management (TAC SCM) scenario provides a unique testbed for studying supply chain management agents. This paper introduces TacTex-05, the champion agent from the 2005 competition, focusing on its ability to adapt to opponent behavior over a series of games. The impact of this adaptivity is examined through both analysis of competition results and controlled experiments.

The TAC SCM tournament is moving into its fourth year. In an effort to track agent progress, we present a benchmark market efficiency comparison for the tournament, in addition to prior measures of agent competency through customer bidding. Using these benchmarks we find statistically significant increases in intratournament market efficiency, whereas agents are generally decreasing in manufacturer market power. We find that agent market share and bid efficiency have increased while the variance of average sales prices has been significantly reduced. Additionally, we test for a statistical relationship between agent profits and the bullwhip effect.

This paper focuses on the Coalition Formation paradigm as a market mechanism. Concretely, Coalition Formation occurs as part of a wider open world and may occur many times during the lifetime of a population of agents. This fact can in some circumstances be exploited by agents to re-use existing partial coalition and social relationships over time to improve Coalition Formation efficiency. The aim of the work is to analyze the dynamics of two concrete rational behaviors (Competitive and Conservative strategies) and, in particular, to investigate how agents in a heterogeneous population cluster together across multiple Coalition Formation episodes and varying tasks. Preliminary resuls are also shown regarding the manner in which playing distinct strategies interact with one another.

In this game, agents will face-off against each other in the ultra-competitive real estate market. Each competitor will act as a real estate agent, working on behalf of clients who need to move into new homes. These clients need to buy a new home as well as sell their current home. The game will test competitors’ technology in two main research areas: preference elicitation and multi-issue negotiation. Each time an agent acquires a new client, it must query the client about its various preferences (e.g. price range, number of bedrooms, etc.) for their new home. Agents then search the listings of the other agents, seeking a possible match. Once found, the agent then engages in negotiations with the selling agent, haggling over various aspects of the deal. Once a house has been purchased, the client’s old house needs to be sold. The objective of the game is to earn the most money. Selling agents earn commissions from sales. Buying agents do not earn commissions, but instead need to maximize the utility of their clients by obtaining a good deal. Satisfied clients are more likely to keep their agent to sell their old house.

In this paper, we investigate the effectiveness of different types of bidding behaviour for trading agents in the Continuous Double Auction (CDA). Specifically, we consider behavioural types that are (expected profit maximising), (targeting a higher profit than neutral) and (trading off profit for a better chance of transacting). For these types, we employ an evolutionary game-theoretic analysis to determine the population dynamics of agents that use them in different types of environments, including dynamic ones with market shocks. From this analysis, we find that given a symmetric demand and supply, agents are most likely to adopt neutral behaviour in static environments, while there tends to be more passive than neutral agents in dynamic ones. Furthermore, when we have asymmetric demand and supply, agents invariably adopt passive behaviour in both static and dynamic environments, though the gain in so doing is considerably smaller than in the symmetric case.

In this paper, we consider the problem of a shop agent negotiating with customers about a bundle of goods or services together with a price. To facilitate the shop agent’s search for mutually beneficial alternative bundles, we develop a method for online learning customers’ preferences, while respecting their privacy. By introducing additional parameters, we represent customers’ highly nonlinear preferences as a linear model. We develop a method for learning the underlying stochastic process of these parameters online. As the conducted computer experiments show, the developed method has a number of advantages: it scales well, the acquired knowledge is robust towards changes in the shop’s pricing strategy, and it performs well even if customers behave strategically.

In this paper, we examine the problem of choosing discriminatory prices for customers with probabilistic valuations and a seller with indistinguishable copies of a good. We show that under certain assumptions this problem can be reduced to the continuous knapsack problem (CKP). We present a new fast -optimal algorithm for solving CKP instances with asymmetric concave reward functions. We also show that our algorithm can be extended beyond the CKP setting to handle pricing problems with overlapping goods (e.g.goods with common components or common resource requirements), rather than indistinguishable goods.

We provide a framework for learning distributions over customer valuations from historical data that are accurate and compatible with our CKP algorithm, and we validate our techniques with experiments on pricing instances derived from the Trading Agent Competition in Supply Chain Management (TAC SCM). Our results confirm that our algorithm converges to an -optimal solution more quickly in practice than an adaptation of a previously proposed greedy heuristic.