A sensing system sometimes requires a complicated optical unit consisting of multiple mirrors, in which case it is important to estimate accurately constitutive parameters of the optical unit to enhance its sensing ca...
详细信息
A sensing system sometimes requires a complicated optical unit consisting of multiple mirrors, in which case it is important to estimate accurately constitutive parameters of the optical unit to enhance its sensing capability. However, the parameters include generally uncertainties since the optical unit cannot avoid the fixing and aligning errors and the manufacturing tolerance of its components. Accordingly, it should construct a projective model of the complicated sensing system accurately and build up an estimation method of tangled parameters. However, it is not easy to estimate complicated constitutive parameters from an accurate model of an optical unit with multiple mirrors, and moreover, they are sometimes changed during operation due to unexpected disturbance or intermittent adjustments such as computer control zoom, auto focus, and mirror relocation. Due to these operational circumstances, it is not easy to take apart components of the assembled system and directly measure the components. Therefore, an indirect and adaptive estimation method, taking all the components into simultaneous consideration without disassembling the sensing system, is needed for calibrating the uncertain and changeable constitutive parameters. In this paper, we propose not only a generalized projective model for an optical sensing system consisting of n-mirrors and a camera with a collecting lens, bur also a learning-based process using the model to estimate recursively the uncertain constitutive parameters of the optical sensing system. We also show its feasibility through a series of calibration of an optical system. (C) 2000 Pattern Recognition Society. Published by Elsevier Science Ltd. All rights reserved.
We propose an efficient alternative to multi-layer perceptron (MLP): two-layer periodic perceptron (PP). We prove then that PP can compute every binary boolean function, we give an efficient learning algorithm for PP ...
详细信息
We propose an efficient alternative to multi-layer perceptron (MLP): two-layer periodic perceptron (PP). We prove then that PP can compute every binary boolean function, we give an efficient learning algorithm for PP and test it on academic and realistic problems. (C) 2000 Elsevier Science B.V. All rights reserved.
In the paper, a novel, fast and accurate artificial neural network is proposed for efficient computer-aided design (CAD) modelling of microstrip discontinuities. The authors lay the goundwork: for their investigation ...
详细信息
In the paper, a novel, fast and accurate artificial neural network is proposed for efficient computer-aided design (CAD) modelling of microstrip discontinuities. The authors lay the goundwork: for their investigation of radial-wavelet neural networks RWNN and their application, to determine the scattering parameters of the circuit under study. Wavelet theory may be exploited in deriving a good initialisation for the neural network, and thus improved convergence of the learning algorithm. The problem of finding a good model is then discussed through solutions offered by radial-wavelet networks trained by Broyden-Fletcher-Goldfarb-Shanno (BFGS) and limited memory BFCS (LBFGS) algorithms. Finally, experimental results, which confirm the validity of the RWNN model, are reported.
This letter presents a novel fuzzy neural network, which is composed of an antecedent network and a consequent network. The antecedent network matches the premises of the fuzzy rules and the consequent network impleme...
详细信息
This letter presents a novel fuzzy neural network, which is composed of an antecedent network and a consequent network. The antecedent network matches the premises of the fuzzy rules and the consequent network implements the consequences of the rules. In the network learning and training phase, a concise and effective algorithm based on the fuzzy hierarchy error approach (FHEA) is proposed to update the parameters of the network. This algorithm is simple to implement and it does not require as many Calculations as some other classic neural network learning algorithms, A model reference adaptive control structure incorporating the proposed fuzzy neural network is studied. Simulation results Of a cart-pole balancing system demonstrate the effectiveness of proposed method.
Capacity control in perceptron decision trees is typically performed by controlling their size. We prove that other quantities can be as relevant to reduce their flexibility and combat overfitting. In particular, we p...
详细信息
Capacity control in perceptron decision trees is typically performed by controlling their size. We prove that other quantities can be as relevant to reduce their flexibility and combat overfitting. In particular, we provide an upper bound on the generalization error which depends both on the size of the tree and on the margin of the decision nodes. So enlarging the margin in perceptron decision trees will reduce the upper bound on generalization error. Based on this analysis, we introduce three new algorithms, which can induce large margin perceptron decision trees. To assess the effect of the large margin bias, OC1 (Journal of Artificial Intelligence Research, 1994, 2, 1-32.) of Murthy, Kasif and Salzberg, a well-known system for inducing perceptron decision trees, is used as the baseline algorithm. An extensive experimental study on real world data showed that all three new algorithms perform better or at least not significantly worse than OC1 on almost every dataset with only one exception. OC1 performed worse than the best margin-based method on every dataset.
This paper discusses neural network-based strategy for reducing the existing errors of fiber-optic gyroscope (FOG), A series-single-layer neural network, which is composed of two single-layer networks in series, is pr...
详细信息
This paper discusses neural network-based strategy for reducing the existing errors of fiber-optic gyroscope (FOG), A series-single-layer neural network, which is composed of two single-layer networks in series, is presented for eliminating random noises. This network has simpler architecture, faster learning speed, and better performance compared to conventional backpropagation (BP) networks. Accordingly, after considering the characteristics of the power law noise in FOG, an advanced learning algorithm is proposed by using the increments of errors in energy function. Furthermore, a radial basis function (RBF) neural network-based method is also posed to evaluate and compensate the temperature drift of FOG. The orthogonal least squares (OLS) algorithm is applied due to its simplicity, high accuracy, and fast learning speed. The simulation results show that the series-single-layer network (SSLN) with the advanced learning algorithm provides a fast and effective way for eliminating different random noises including stable and unstable noises existing in FOG, and the RBF network-based method offers a powerful and successful procedure for evaluating and compensating the temperature drift.
This paper presents an extension of the standard backpropagation algorithm (SBP). The proposed learning algorithm is based on the fuzzy integral of Sugeno and thus called fuzzy backpropagation (FBP) algorithm. Necessa...
详细信息
This paper presents an extension of the standard backpropagation algorithm (SBP). The proposed learning algorithm is based on the fuzzy integral of Sugeno and thus called fuzzy backpropagation (FBP) algorithm. Necessary and sufficient conditions for convergence of FBP algorithm for single-output networks in case of single- and multiple-training patterns are proved. A computer simulation illustrates and confirms the theoretical results, FBP algorithm shows considerably greater convergence rate compared to SEP algorithm. Other advantages of FBP algorithm are that it reaches forward to the target value without oscillations, requires no assumptions about probability distribution and independence of input data. The convergence conditions enable training by automation of weights tuning process (quasi-unsupervised learning) pointing out the interval where the target value belongs to. This supports acquisition of implicit knowledge and ensures wide application, e.g. for creation of adaptable user interfaces, assessment of products, intelligent data analysis, etc. (C) 2000 Elsevier Science B.V. All rights reserved.
The conformality of the self-organizing network is studied in this work. We use multi-dimensional deformation analyses to interpret the self-organizing mapping, It can be shown that this mapping is quasi-conformal wit...
详细信息
The conformality of the self-organizing network is studied in this work. We use multi-dimensional deformation analyses to interpret the self-organizing mapping, It can be shown that this mapping is quasi-conformal with a convergent deformation bound. Based on analyses, a deformation measure and a non-conformality measure are derived to indicate the evolution status of the network. These measures can serve as new criteria to evolve the network. We test these measures with simulations on surface mapping problems. (C) 2000 Elsevier Science B.V. All rights reserved.
The inversion of a neural network is a process of computing inputs that produce a given target when fed into the neural network. The inversion algorithm of crisp neural networks is based on the gradient descent search...
详细信息
The inversion of a neural network is a process of computing inputs that produce a given target when fed into the neural network. The inversion algorithm of crisp neural networks is based on the gradient descent search in which a candidate inverse is iteratively refined to decrease the error between its output and the target. In this paper, me derive an inversion algorithm of fuzzified neural networks from that of crisp neural networks. First, we present a framework of learning algorithms of fuzzified neural networks and introduce the idea of adjusting schemes for fuzzy variables. Next, we derive the inversion algorithm of fuzzified neural networks by applying the adjusting scheme for fuzzy variables to total inputs in the input layer, Finally, we make three experiments on the parity-three problem;we examine the effect of the size of training sets on the inversion and investigate how the fuzziness of inputs and targets of training sets affects the inversion.
Image compression is an important area of multimedia investigation and neural network methods have attracted more and more attentions for using in image coding. Recently a random neural network model, which has the so...
详细信息
ISBN:
(纸本)0819438758
Image compression is an important area of multimedia investigation and neural network methods have attracted more and more attentions for using in image coding. Recently a random neural network model, which has the solutions with product form in steady state (i.e. the steady state probability distribution of network can always be expressed as the product of the probabilities of the states of each neuron) on some conditions, was brought forward. Among the diverse random neural network models, the feed-forward one is very practicable because its solutions exist and are unique. In this paper, a new learning method for feed-forward random neural network, which can be implemented easier than the learning algorithm of the RNN presented by Gelenbe, was presented. Using the new learning formulas we developed, we designed a new image coding method, which applies the random neural network method in classical DCT-based coding framework. The experimental results show that our new method could gain a lot in PSNR (1 similar to 2dB) compared with standard neural network coding methods. In conclusion, we stated that the DCT-based image compression method using random neural network is an efficient algorithm for image coding.
暂无评论