Catálogo de publicaciones - libros

This work describes the development and results on an agent architecture devoted to the planning and scheduling of passenger trips by using the contract-net protocol as base coordination mechanism. The architecture named MADARP, has been implemented over Jade, and provides a set of base agents that perform the basic interface, planning and support services, which can be extended to tackle specific passenger transport conditions and scenarios. In particular, this paper will focus on the planning agents, their coordination mechanism and its performance in a distributed scenario. The agent use allows easily adapting the architecture to different assigning and scheduling models. This work presents the results of a performance test, analyzing two different rates for the requests’ arrivals under a distributed scenario with diverse number of hosts.

We propose two new models for view-independent face recognition, which lies under the category of multiview approaches. We use the so-called “mixture of experts” (MOE) in which, the problem space is divided into several subspaces for the experts, and then the outputs of experts are combined by a gating network to form the final output. Basically, our focus is on the way that the face space is partitioned by MOE. In our first model, experts of MOE structure are not biased in any way to prefer one class of faces to another, in other words, the gating network learns a partition of input face space and trusts one expert in each of these partitions; we call this method “self-directed partitioning”. In our second model, we attempt to direct the experts to specialize in predetermined areas of face space by developing teacher-directed learning methods for MOE. In this model, by including teacher information about the pose of input face image in the training phase of networks, each expert is directed to learn faces of a specific pose class, so referred to as “teacher-directed partitioning”. Thus, in our second model, instead of allowing the MOE to partition the face space on its own way, it is quantized according to a number of predetermined views and MOE is trained to adapt to such space partitioning. The experimental results support our claim that directing the mixture of experts to a predetermined partitioning of face space is a more beneficial way of using MOE for view-independent face recognition.

For better performance and to avoid member service annoyance that results due to joining-clients’ waiting durations and time-outs when there are more than one clients wanting to join concurrently for FTTH-based Broadcasting Mini-system’s service, this paper proposes a more efficient and better performing Overlay Tree Building Control Protocol by modifying and extending the basic mechanisms building the conventional TBCP. The modified-TBCP (TBCP) proposed is performance-effective mechanism since it considers the case of how fast will children, concurrently, find and join new parents when paths to existing parents are broken. Besides utilizing partial topology information, TBCP also does a LAN-out-degree-check. If the selected child-parent-pair falls under the same LAN, that selected parent does not change the out-degree status. The performance comparison, in terms of Overlay-Connection-Count and Latency against Group-Size-Growth, between the proposed TBCP and the traditional TBCP is done through simulations and the results conclude in favour of the proposed TBCP.

We study the problem of allocating a single divisible resource to strategic agents in communication networks. For the setting of inelastic supply we formulate conditions for the efficiency of Nash equilibria in a resulting resource allocation game. Then we answer the question how to design a mechanism which yields efficient allocations and analyze revenues generated by the proposed class of mechanisms. Finally, we propose a distributed resource allocation algorithm and prove its convergence.

P2P network has been emerging as a very attractive content distribution mechanism. Nevertheless flooding based applications and systems with distribution hash tables based model, which are already developed, are having difficulty in the development and the application because they do not satisfy many issues such as privacy protection of end-user. In this study we propose neighbor-set based privacy protection model. The proposed technique is based on the typical distributed hash tables model, chord, and was designed to be protected from attack by securing anonymity to participating nodes. Proposed protecting scheme and how to solve the problem of load balancing caused in the process will be explained.

In this work we assume that there is an agent in an unknown environment (domain). This agent has some predefined actions and it can perceive its current state in the environment completely. The mission of this agent is to fulfill the tasks (goals) that are often assigned to it as fast as it can. Acting has lots of cost, and usually planning and simulating the environment can reduce this cost. In this paper we address a new approach for incremental induction of probabilistic planning operators, from this environment while the agent tries to reach to its current goals. It should be noted that there have been some works related to incremental induction of deterministic planning operators and batch learning of probabilistic planning operators, but the problem of incremental induction of probabilistic planning operators has not been studied yet. We also address some trade offs such as exploration (for better learning of stochastic operators, acting) and exploitation (for fast discovery of goals, planning), and we explain that a good decision in these trade offs is dependant on the stability and accuracy of the learned planning operators.

The proposed Competitive Contract Net Protocol has been designed to facilitate a flexible cooperation in competitive multi-agent environments and to support automated or semi-automated negotiations in competitive domains. The protocol is based on FIPA standards. The protocol covers not only the phase of contracting the commitments, but also allows for a decommitment negotiation and contract termination. Thus, it consists of three phases: (i) a contracting phase, where conditions of agreement are concluded, (ii) an optional decommitment phase, where contract may be breached, and (iii) a contract termination phase, where the compliance with the concluded contract conditions is evaluated. Both the decommitment and non-compliance are bounded with penalties which measurably ensure a compliance with the commitments, but also allow an opportunistic behaviour of the agents at some price.

The agent-oriented (AO) methodology is an effective means for constructing complex systems. Despite a great deal of research, a number of challenges still exist before making agent-based computing a widely accepted paradigm in software engineering practice. In order to solve the problem of “a variety in number, difficult to use”, the paper presents a hierarchical development architecture (HDA) for customizing a new AO methodology according to the given project. Based on the HDA, the developer can extract meta models from existing AO methods to assemble a new approach, much like developers building applications from third party off-the-shelf components. To exemplify its feasibility and effectiveness, the construction of C4I system is presented as a case study.

The increasing number of Semantic Web applications that work with ontologies implies an increased need for building ontological knowledge bases. In order to improve ontologies during their development as well as to allow applications to be experimentally evaluated prior to their complete implementation and deployment, ontology bases must be filled with experimental data (i.e., instance ontologies), which can be used to evaluate methods used for information processing. We describe several approaches together with a method of building an ontological knowledge base for purposes of experimentation with Semantic Web applications. We also discuss characteristics and suitability of particular approaches to the development of experimental ontological knowledge bases.

Existing item bank systems present a variety of assessments for data management and integration with individual learning evaluation systems. However, as the data in these established item bank systems do not include semantics, the bank systems cannot analyze implications and perform accurate searches such as for synonymous words. Therefore, both learners and teachers can access only simple text data using the item bank system and often waste time checking unnecessary search results and extracting information from data repeatedly. Moreover, since there is no clear definition of the relationship between items and teachers or learners, or between data and units, it is difficult to use and share extra-item information. In order to solve these problems, this research describes the definition of conception and organization by constructing the Ontology of an E-Item Bank system using OWL, which is a form of semantic web technology. Furthermore, on the basis of Ontology, OWL metadata, individuals, are built, semantic factors are then extracted using OWLJessKB. We do not only use this extracted semantic factors as a fact, but also domain rules using SWRL for the JESS inferencing engine. As such, it is possible to make inferences and provide reasoning in web data structures. As a result, it is possible to combine all the data of the E-Item bank system and facilitate understanding of the meanings of concepts by the computer. In addition, a search service with inference can be applied to education and can lead to cooperative study between teachers and students.