F. J. Fernández

Multiobjective evolutionary optimization of the size, shape, and position parameters of radial basis function networks for function approximation

This paper presents a multiobjective evolutionary algorithm to optimize radial basis function neural networks (RBFNNs) in order to approach target functions from a set of input-output pairs. The procedure allows the application of heuristics to improve the solution of the problem at hand by including some new genetic operators in the evolutionary process. These new operators are based on two well-known matrix transformations: singular value decomposition (SVD) and orthogonal least squares (OLS), which have been used to define new mutation operators that produce local or global modifications in the radial basis functions (RBFs) of the networks (the individuals in the population in the evolutionary procedure). After analyzing the efficiency of the different operators, we have shown that the global mutation operators yield an improved procedure to adjust the parameters of the RBFNNs.

New fluorescent pH sensors based on covalently linkable PET rhodamines.

A new class of rhodamines for the application as indicator dyes in fluorescent pH sensors is presented. Their pH-sensitivity derives from photoinduced electron transfer between non-protonated amino groups and the excited chromophore which results in effective fluorescence quenching at increasing pH. The new indicator class carries a pentafluorophenyl group at the 9-position of the xanthene core where other rhodamines bear 2-carboxyphenyl substituents instead. The pentafluorophenyl group is used for covalent coupling to sensor matrices by “click” reaction with mercapto groups. Photophysical properties are similar to “classical” rhodamines carrying 2′-carboxy groups. pH sensors have been prepared with two different matrix materials, silica gel and poly(2-hydroxyethylmethacrylate). Both sensors show high luminescence brightness (absolute fluorescence quantum yield ΦF≈0.6) and high pH-sensitivity at pH 5–7 which makes them suitable for monitoring biotechnological samples. To underline practical applicability, a dually lifetime referenced sensor containing Cr(III)-doped Al2O3 as reference material is presented.

A two-stage approach to self-learning direct fuzzy controllers

In this paper, a novel approach is presented to fine tune a direct fuzzy controller based on very limited information on the nonlinear plant to be controlled. Without any off-line pretraining, the algorithm achieves very high control performance through a two-stage algorithm. In the first stage, coarse tuning of the fuzzy rules (both rule consequents and membership functions of the premises) is accomplished using the sign of the dependency of the plant output with respect to the control signal and an overall analysis of the main operating regions. In stage two, fine tuning of the fuzzy rules is achieved based on the controller output error using a gradient-based method. The enhanced features of the proposed algorithm are demonstrated by various simulation examples.

Multidimensional and multideme genetic algorithms for the construction of fuzzy systems

Abstract In this paper, a real-coded genetic algorithm (GA) is proposed capable of simultaneously optimizing the structure of a system (number of inputs, membership functions and rules) and tuning the parameters that define the fuzzy system. A multideme GA system is used in which various fuzzy systems with different numbers of input variables and with different structures are jointly optimized. Communication between the different demes is established by the migration of individuals presenting a difference in the dimensionality of the input space of a particular variable. We also propose coding by means of multidimensional matrices of the fuzzy rules such that the neighbourhood properties are not destroyed by forcing it into a linear chromosome. The effectiveness of the proposed approach is verified by a variety of simulation examples and is compared with other fuzzy, neuro-fuzzy and fuzzy–genetic approaches in terms of the root-mean-square error (RMSE).

A new approach for the design of fuzzy controllers in real time

This paper presents a new methodology to achieve real time self tuning and self-learning in fuzzy controllers. The advantage of this approach is that it only requires qualitative information about the plant to be controlled, in terms of the monotony presented by the output with respect to the control signal and delays of the plant. Thus, it is capable of controlling highly nonlinear systems, in a pseudo-optimum way, even when these are time variable. Control is achieved by means of two auxiliary systems: the first one is responsible for adapting the consequences of the main controller to minimize the error arising at the plant output, while the second auxiliary system compiles real input/output data obtained from the plant. The system then learns from these data, adapting both the consequences of the rules and the parameters that define the membership functions, taking into account, not the current state of the plant but rather the global identification performed.

A new methodology to obtain fuzzy systems autonomously from training data

This paper presents an approach to obtain a fuzzy system automatically from numerical data. The identification of the fuzzy system structure (number of rules and membership functions in each input variable) and the optimization of the parameters defining it are performed jointly. Starting from an initially simple fuzzy system, the numbers of membership functions in the input domain and of rules are adapted in order to reduce the approximation error. This method has the advantage that it does not require the human experts assistance since the input-output characteristics of the fuzzy system and its structure are obtained from the training examples.

An efficient OS support for communication on Linux clusters

A communication layer is proposed that, besides improving communication performance on clusters of PCs, by reducing the latencies and increasing the bandwidth figures even for short messages, also meets other requirements such as multiprogramming, portability, protection against corrupted programs, reliable message delivery, direct access to the network for all applications, etc. Instead of removing the operating system kernel from the critical path and creating a user-level network interface, our aim was to optimize the operating system support to provide reliable and efficient network software, avoiding the TCP/IP protocol stack. The communication system was tested in a cluster of PCs with Linux OS and interconnected with Fast Ethernet. The performance figures obtained define the best situation that can be attained without modifying the device drivers or using a user-level network interface approach.

A VLSI Approach to the Implementation of Additive and Shunting Neural Networks

Biologically inspired VLSI circuits are proposed which can be particularized to approximate the real-time dynamics of either additive or shunting neural models. Analog inputs to these circuits are represented by short spikes and, both, their transient and steady-state behaviours depend only on process-independent local ratios. The paper includes simulation results and experimental measures of a CMOS prototype, which illustrate the utility and the feasibility of the proposed VLSI approach.

An Approach to Isolated Word Recognition Using Multilayer Perceptrons

Neural networks offer the potential of providing massive parallelism, adaptation, and new algorithm approaches to speech recognition. In this communication, we show a new approach to face the problem of speaker-independent isolated word recognition with the Multilayer Perceptron (MLP), trained with Backpropagation algorithm. This approach lies in a preprocessing similar to that used for Kohonen Networks, thus in the context of unsupervised learning, which allows to overcome the temporal alignment problem of word samples and to reduce the number of neurons in the MLPs. As a preliminary result, the performances of MLPs for recognizing sequences of vowels in isolated words, after learning with samples of isolated vowels, are presented.

Simulation and hardware implementation of competitive learning neural networks

One of the main research topics within neuron-like networks is related to learning techniques. Competitive learning has got an special interest among them, because a great network automation is achieved with it, ie, autonomously and without explicit indication of the correct output patterns, the network extracts general features that can be used in order to cluster a set of patterns. In this paper, after giving a brief overview about learning procedures, the most peculiar characteristics of competitive learning are pointed out, and the different ways of implementing neuron-like networks are quoted, describing as an implementation instance our present project of hardware construction of a neural chip to be included in a coprocessor board with competitive learning.