José Manuel Molina López

A fuzzy model of reputation in multi-agent systems

Agents are intended to interact in open systems where the knowledge about others (reputation) is incomplete and uncertain. Also, this knowledge about other agents is subjective since it involves how much the services provided by other agents suit the user’s expectations. So reputation is not considered a crisp value since the knowledge we have about other agents is incomplete, vague and uncertain. Our proposal adopts a distributed approach to reputation management where reputation is built through the own past experience and the references (or recommendations) received from other agents. The system presented in this paper is driven to personalization and efficiency. Personalization is achieved representing explicitly a fuzzy reputation of other agents. Therefore, the interaction among agents will be closer to human way of thinking. The system deals with changes in fuzzy reputation values and with the exchange of messages driven to a cooperation among agents. As a result of the fuzzy aggregation, we’ll expect the agents to be grouped in clusters of common preferences. Most of the required matchmaking would be perform inside the clusters leading to a very efficient distribution of communications The model is implemented and tested using Fril. It is a prolog-like logic programming language incorporating fuzzy and probabilistic uncertainty.

Effective Evolutionary Algorithms for Many-Specifications Attainment: Application to Air Traffic Control Tracking Filters

This paper addresses a real-world engineering design requiring the application of effective and global optimization techniques. The problem it deals with is the design of nonlinear tracking filters under up to several hundreds of performance specifications. The suitability of different evolutionary computation techniques for solving multiobjective problems is explored, contrasting the performance achieved with recent multiobjective evolutionary algorithm (MOEAs) proposals and different aggregation schemes. In particular, a new scheme is proposed to build a fitness function based on an operator that selects worst cases of multiple specifications in different situations. They have been evaluated in the design of an air traffic control (ATC) tracking filter that should accomplish a specific normative with 264 specifications. Results show their performance in terms of effectiveness and computational load, comparing their capability to scale the problem with respect to problem size.

Recognizing Human Activities from Sensors Using Hidden Markov Models Constructed by Feature Selection Techniques

In this paper a method for selecting features for Human Activity Recognition from sensors is presented. Using a large feature set that contains features that may describe the activities to recognize, Best First Search and Genetic Algorithms are employed to select the feature subset that maximizes the accuracy of a Hidden Markov Model generated from the subset. A comparative of the proposed techniques is presented to demonstrate their performance building Hidden Markov Models to classify different human activities using video sensors.

MIJ2K: Enhanced video transmission based on conditional replenishment of JPEG2000 tiles with motion compensation

A video compressed as a sequence of JPEG2000 images can achieve the scalability, flexibility, and accessibility that is lacking in current predictive motion-compensated video coding standards. However, streaming JPEG2000-based sequences would consume considerably more bandwidth. With the aim of solving this problem, this paper describes a new patent pending method, called MIJ2K. MIJ2K reduces the inter-frame redundancy present in common JPEG2000 sequences (also called MJP2). We apply a real-time motion detection system to perform conditional tile replenishment. This will significantly reduce the bit rate necessary to transmit JPEG2000 video sequences, also improving their quality. The MIJ2K technique can be used both to improve JPEG2000-based real-time video streaming services or as a new codec for video storage. MIJ2K relies on a fast motion compensation technique, especially designed for real-time video streaming purposes. In particular, we propose transmitting only the tiles that change in each JPEG2000 frame. This paper describes and evaluates the method proposed for real-time tile change detection, as well as the overall MIJ2K architecture. We compare MIJ2K against other intra-frame codecs, like standard Motion JPEG2000, Motion JPEG, and the latest H.264-Intra, comparing performance in terms of compression ratio and video quality, measured by standard peak signal-to-noise ratio, structural similarity and visual quality metric metrics.

MIJ2K Optimization using evolutionary multiobjective optimization algorithms

This paper deals with the multiobjective definition of video compression and its optimization. The optimization will be done using NSGA-II, a well-tested and highly accurate algorithm with a high convergence speed developed for solving multiobjective problems. Video compression is defined as a problem including two competing objectives. We try to find a set of optimal, so-called Pareto-optimal solutions, instead of a single optimal solution. The two competing objectives are quality and compression ratio maximization. The optimization will be achieved using a new patent pending codec, called MIJ2K, also outlined in this paper. Video will be compressed with the MIJ2K codec applied to some classical videos used for performance measurement, selected from the Xiph.org Foundation repository. The result of the optimization will be a set of near-optimal encoder parameters. We also present the convergence of NSGA-II with different encoder parameters and discuss the suitability of MOEAs as opposed to classical search-based techniques in this field.