Neural Networks for Signal Processing IX: Proceedings of the 1999 IEEE Signal Processing Society Workshop (Cat. No.98TH8468)

25-25 Aug. 1999

Filter Results

Displaying Results 1 - 25 of 61
  • Neural Networks for Signal Processing IX: Proceedings of the 1999 IEEE Signal Processing Society Workshop (Cat. No.98TH8468)

    Publication Year: 1999
    Request permission for commercial reuse | PDF file iconPDF (218 KB)
    Freely Available from IEEE
  • Author index

    Publication Year: 1999, Page(s):565 - 566
    Request permission for commercial reuse | PDF file iconPDF (61 KB)
    Freely Available from IEEE
  • Why a nonlinear solution for a linear problem? [channel equalization]

    Publication Year: 1999, Page(s):157 - 165
    Cited by:  Papers (4)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (440 KB)

    We emphasize a key point that when there is noise in the system, even if the system is linear, a nonlinear solution is more desirable. We derive a simple expression that shows that for a linear regression model, the logistic nonlinearity will be the natural match for modeling posterior class probabilities, and that the steepness of this logistic function is inversely proportional to the level of n... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Partial likelihood methods for probability density estimation

    Publication Year: 1999, Page(s):147 - 156
    Cited by:  Papers (4)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (428 KB)

    Partial likelihood (PL) establishes a sufficiently general framework to develop and study statistical properties of nonlinear techniques in signal processing. Adah et al. (1997), present the theorem by which the fundamental information-theoretic relationship for learning the PL cost, the equivalence of likelihood maximisation and relative entropy minimization, is established. In this paper, we ref... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A spiking neural network architecture for nonlinear function approximation

    Publication Year: 1999, Page(s):139 - 146
    Cited by:  Papers (3)  |  Patents (2)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (268 KB)

    Multilayer perceptrons have received much attention due to their universal approximation capabilities. Normally such models use real valued signals, although they are loosely based on biological neuronal networks which encode signals using spike trains. Spiking neural networks are of interest from both a biological point of view, but also from a method of robust signalling in particularly noisy or... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A control theory formulation for random variate generation

    Publication Year: 1999, Page(s):132 - 138
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (264 KB)

    The need to simulate complex systems in a Monte Carlo manner necessitates efficient methods for generating random variates. We propose a method for random variate generation. The method is based on a control theory formulation. We use a cascade structure consisting of a neural network “controller” and a density estimator (“plant”). The neural network “controller&rdquo... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • On training piecewise linear networks

    Publication Year: 1999, Page(s):122 - 131
    Cited by:  Papers (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (380 KB)

    Piecewise-linear (PWL) neural networks are networks with piecewise-linear node functions. They have attractive features, such as speed of training and amenability to digital VISI implementation. The paper presents an algorithm for training PWL networks. The algorithm is general in that it can be used for the optimization of general PWL functions. It is based on moving from one linear region to the... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • On optimal data split for generalization estimation and model selection

    Publication Year: 1999, Page(s):225 - 234
    Cited by:  Papers (6)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (456 KB)

    The paper is concerned with studying the very different behavior of the two data splits using hold-out cross-validation, K-fold cross-validation and randomized permutation cross-validation. First we describe the theoretical basics of various cross-validation techniques with the purpose of reliably estimating the generalization error and optimizing the model structure. The paper deals with the simp... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • View-based 3D object recognition with support vector machines

    Publication Year: 1999, Page(s):77 - 84
    Cited by:  Papers (29)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (412 KB)

    Support vector machines have demonstrated excellent results in pattern recognition tasks and 3D object recognition. We confirm some of the results in 3D object recognition and compare it to other object recognition systems. We use different pixel-level representations to perform the experiments, while we extend the setting to the more challenging and practical case when only a limited number of vi... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Learning Gestalt of surfaces in natural scenes

    Publication Year: 1999, Page(s):380 - 389
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (520 KB)

    We develop a computational model for scenes with surfaces that have rough and non-smooth small-scale structure but with a perceived global (larger-scale) geometric form. Examples include grass and meadow, surfaces textured with sand-paper, natural scenes having rough texture such as the skin of crocodile, pine cones, a field of sea urchins, forests, ripples and waves on water surfaces, etc. Anothe... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Some analytical results on critic-driven ensemble classification

    Publication Year: 1999, Page(s):253 - 262
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (428 KB)

    We (1999) proposed a framework for ensemble classification wherein auxiliary networks, dubbed critics, are used to provide reliability information on the ensemble's individual classifiers/experts. We showed experimentally that critic-driven combining schemes extend the applicability of ensemble methods by overcoming the usual requirement that the individual classifier error rate p must be less tha... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • An adaptive metric learning procedure for reconfigurable facial signature authentication

    Publication Year: 1999, Page(s):409 - 418
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (380 KB)

    We present an adaptive metric learning procedure with improved generalization of missing training data for facial signature recognition for use in a smart card system. The conventional learning models suffer from degraded recognition rate due to poor estimation of the margin of a decision boundary. Our model employs an image synthesis method to represent missing patterns of unknown classes by usin... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • General statistical inference by an approximate application of the maximum entropy principle

    Publication Year: 1999, Page(s):112 - 121
    Cited by:  Papers (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (492 KB)

    We propose a learning method for building a general statistical inference engine, operating on discrete feature spaces. Such a model allows inference on any feature given values for the other features (or for a feature subset). Bayesian networks (BNs) are versatile tools that possess this inference capability. However, while the BN's explicit representation of conditional independencies is informa... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Sizing of the multilayer perceptron via modular networks

    Publication Year: 1999, Page(s):215 - 224
    Cited by:  Papers (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (404 KB)

    A fast method for sizing the multilayer perceptron is proposed. The principal assumption is that a modular network with the same theoretical pattern storage as the multilayer perceptron has the same training error. This assumption is analyzed for the case of random patterns. Using several benchmark datasets, the validity of the approach is demonstrated View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Robust machine fault detection with independent component analysis and support vector data description

    Publication Year: 1999, Page(s):67 - 76
    Cited by:  Papers (13)  |  Patents (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (428 KB)

    We propose an approach to fault detection in rotating mechanical machines: fusion of multichannel measurements of machine vibration using independent component analysis (ICA), followed by a description of the admissible domain (part of the feature space indicative of normal machine operation) with a support vector domain description (SVDD) method. The SVDD method enables the determination of an ar... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Efficient approximation of a neural filter for quantum noise removal in X-ray images

    Publication Year: 1999, Page(s):370 - 379
    Cited by:  Papers (5)  |  Patents (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (488 KB)

    An efficient filter approximating the neural filter (NF) trained to remove quantum noise from X-ray images has been realized. A novel analysis method is proposed for making the characteristics of the trained NF clear. It analyses a nonlinear system with plural inputs by using its outputs when the specific input signals are fed to it. The experimental results have demonstrated that the approximate ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Time series classification using adaptive dynamic targets

    Publication Year: 1999, Page(s):243 - 252
    Cited by:  Papers (2)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (396 KB)

    To train a classifier with supervised learning appropriate targets have to be provided. In the case of time series this can be complicated if there is only one target for the whole time series, but the learning algorithm needs a target at each time step. In the paper a technique is introduced, which is able to provide appropriate targets at each time step. As a result of this technique the impact ... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Selective image smoothing via dyadic wavelet-based conduction equation

    Publication Year: 1999, Page(s):400 - 408
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (412 KB)

    We propose a new dyadic wavelet-based conduction approach for selective image smoothing. In our approach, a nonlinear conductivity function is considered in the wavelet-based function decomposition and reconstruction process. Since the proposed approach does not require one to solve a PDE, it is therefore more efficient and accurate than the conventional nonlinear diffusion/conduction-based method... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Near-optimal flight load synthesis using neural nets

    Publication Year: 1999, Page(s):535 - 544
    Cited by:  Papers (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (368 KB)

    This paper describes the use of neural networks for near-optimal helicopter flight load synthesis (FLS), which is the process of estimating mechanical loads during helicopter flight, using cockpit measurements. First, modular neural networks are used to develop statistical signal models of the cockpit measurements as a function of the loads. Then Cramer-Rao maximum a-posteriori bounds on the mean-... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Adaptive Web caching using logistic regression

    Publication Year: 1999, Page(s):515 - 524
    Cited by:  Papers (1)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (468 KB)

    A fundamental understanding of Web access patterns is necessary before we can tackle cache performance issues. Due to the extremely dynamic nature of the Web, any techniques we employ must be able to adapt to changing conditions on the Web. We introduce a near-optimal cache algorithm, based on complete a priori knowledge of future Web access. In reality, such knowledge is unavailable. To enable us... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Hierarchy of probabilistic principal component subspaces for data mining

    Publication Year: 1999, Page(s):497 - 506
    Cited by:  Papers (2)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (612 KB)

    Visual exploration has proven to be a powerful tool for multivariate data mining. Most visualization algorithms aim to find a projection from the data space down to a visually perceivable rendering space. To reveal all of the interesting aspects of complex data sets living in a high-dimensional space, a hierarchical visualization algorithm is introduced which allows the complete data set to be vis... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Discrimination of cylinders with different wall thicknesses using neural networks and simulated dolphin sonar signals

    Publication Year: 1999, Page(s):477 - 486
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (500 KB)

    This paper describes a method integrating neural networks into a system for recognizing underwater objects. The system is based on a combination of simulated dolphin sonar signals, simulated auditory filters and artificial neural networks. The system is tested on a cylinder wall thickness difference experiment and demonstrates high accuracy for small wall thickness differences. Results from the ex... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Multi-channel piecewise selective averaging of cognitive evoked potentials with variable latency

    Publication Year: 1999, Page(s):459 - 467
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (520 KB)

    This work is about the development of an alternative way of averaging evoked potentials (EP) of cognitive activities. Since the main assumption of invariant waveforms time locked to the eliciting events does not hold for cognitive EPs, averaging results in distorted estimates. The authors' alternative selective averaging finds similar subsequences of fixed length with variable latency which are co... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Weight estimation for the learning of modular perceptron networks

    Publication Year: 1999, Page(s):103 - 111
    Cited by:  Papers (2)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (344 KB)

    We propose a weight estimation method for feedforward neural networks. The proposed method includes two steps: (1) weight vector orientation estimation and (2) weight vector length estimation, such that the iteration learning of modular perceptron networks (MPN) can be reduced. We have applied the proposed method to the divide-and-conquer learning (DCL) for two-spiral problems (TPS) on an MPN. The... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Model combination and weight selection criteria for speaker verification

    Publication Year: 1999, Page(s):439 - 448
    Cited by:  Papers (2)  |  Patents (2)
    Request permission for commercial reuse | Click to expandAbstract | PDF file iconPDF (428 KB)

    We focus on the score combination for three separate modeling approaches as applied to text-dependent speaker verification. The modeling methods that are evaluated consist of the neural tree network (NTN), hidden Markov model (HMM), and dynamic time warping (DTW). One of the main challenges in combining scores of several models is how to select the weight for each model. One method is to use equal... View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.