Catálogo de publicaciones - libros

This paper presents a dynamic programming algorithm to draw optimal intermodal freight routing with regard to international logistics of container cargo for export and import. This study looks into the characteristics of intermodal transport using two or more modes, and presents a Weighted Constrained Shortest Path Problem (WCSPP) model. This study draws a Pareto optimal solution that can simultaneously meet two objective functions by applying the Label Setting algorithm, one of the Dynamic Programming algorithms, after setting feasible area using the objective function values drawn in the model. To improve the algorithm performance, pruning rules have also been presented. The algorithm is applied to real transport paths from Busan to Rotterdam. This study quantitatively measures the savings effect of transport cost and time by comparing single transport modes with existing intermodal transport paths. Lastly, this study applies the multiple Pareto optimal solutions drawn, to a mathematical model and MADM model, and compares the two evaluation methods as a means to evaluate the solutions.

The article presents an optimization method of combustion process in a power boiler. This solution is based on the artificial immune systems theory. A layered optimization system is used, to minimize CO and NOx emission. Immune inspired optimizer SILO is implemented in each of three units of Ostroleka Power Plant (Poland). The results from this implementation are presented. They confirm that presented solution is effective and usable in practice.

This paper presents a technique for dynamically determining search spaces in order to enable sensor exploration during autonomous search and tracking (SAT) missions. In particular, marine search and rescue scenarios are considered, highlighting the need for exploration during SAT. A comprehensive method which is independent of search space representation is introduced, based on exploration frontiers and reachable set analysis. The advantage of the technique is that recursive Bayesian estimation can be performed indefinitely, without loss of information. Numerical results involving multiple search vehicles and multiple targets demonstrate the efficacy of the approach for coordinated SAT. These examples also highlight the added benefit for human mission planners resulting from the technique’s simplification of the search space allocation task.

This paper presented an image reconstruction method for a capsule endoscope. The proposed method constructs a 3–D model of the intestine using massive images obtained from the capsule endoscope. It merges all images and yields a complete 3-D model of the intestine. This 3-D model is reformed as a 2-D plane image showing the inner side of the entire intestine. The proposed image composition has been evaluated using the OpenGL 3-D simulator. The composite image provides an easy-to-understand view for examining the intestine. In addition, it provides fast track-and-check diagnosis using the3-D model implementation.

Collaborative filtering (CF) technique has been widely used in recommending items of interest to users based on social relationships. The notion of trust is emerging as an important facet of relationships in social networks. In this paper, we present an improved mechanism to the standard CF techniques by incorporating trust into CF recommendation process. We derive the trust score directly from the user rating data and exploit the trust propagation in the trust web. The overall performance of our trust-based recommender system is presented and favorably compared to other approaches.

A hybrid recommendation method is presented in this paper. Its main goal is to improve recommendation recall maintaining high recommendation precision and adaptive ability. The formal model is used to define the method and to analyze how the measures known from traditional Information Retrieval may be adapted to recommendation. The presented theorems show that the method is able to adapt to changing user’s needs and achieve the maximal effectiveness if the component methods work properly.

Although combinatorial auction has been studied extensively, it is difficult to apply the existing results to a problem with minimal resource requirements. In this paper, we consider a combinatorial auction problem in which an auctioneer wants to acquire resources from a set of bidders to process the tasks on hand. Each task requires a minimal set of resources for executing the operations. Each bidder owns a set of resources to bid for the tasks. The problem is to determine the resource assignment to minimize the total cost to perform the tasks. The main results include: (1) a problem formulation for combinatorial auction with minimal resource requirements; (2) a solution methodology based on Lagrangian relaxation; (3) an economic interpretation and a proposed structure for implementing our solution algorithms.

The growing complexity of communication networks and their associated information overhead have made network management considerably difficult. This paper presents a novel Network Management Scheme based on the novel concept of Active Information Resources (AIRs). Many types of information are distributed in the complex network, and they are changed dynamically. Under the AIR scheme, each piece of information in a network is activated as an intelligent agent: an I-AIR. An I-AIR has knowledge and functionality related to its information. The I-AIRs autonomously detect run-time operational obstacles occurring in the network system and specify the failures’ causes to the network administrator with their cooperation. Thereby, some network management tasks are supported. The proposed prototype system (AIR-NMS) was implemented. Experimental results indicate that it markedly reduces the network administrator workload, compared to conventional network management methods.

The agent-based systems have been designed and developed using the latest agent technologies. However, the design and the debugging of these systems contain some difficult problems due to the situational and nondeterministic nature of the agents, and the effective design-supporting technologies have not been proposed so far. In order to make efficient design process of agent system, we propose an interactive development method of agent system based on the agent-repository-based multiagent framework which focuses on an essential feature of agent system design, i.e., the reuse of existing agents stored in the agent repository. In this paper, we propose an Interactive Design Environment of Agent system called IDEA and demonstrate the effectiveness of the proposed method.

This paper proposes a new artificial intelligence framework that solves knapsack related problems (a class of complex combinatorial optimization problems). This framework, which is pseudo-parallel and stochastic, uses an agent based system to approximately solve the optimization problem. The system consists of active agents interacting in real time. They mimic the behavior of the parameters of the optimization problem while being individually driven by their own parameters, decision process, and fitness assessment. The application of the framework to the two-dimensional guillotine bin packing problem demonstrates its effectiveness both in terms of solution quality and run time.