Notification:
We are currently experiencing intermittent issues impacting performance. We apologize for the inconvenience.
By Topic

Neural Networks, 2004. Proceedings. 2004 IEEE International Joint Conference on

Date 25-29 July 2004

Go

Filter Results

Displaying Results 1 - 25 of 184
  • Single categorizing and learning module for temporal sequences

    Publication Year: 2004 , Page(s): 2977 - 2982 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (724 KB) |  | HTML iconHTML  

    Modifications of an existing neural network called categorizing and learning module (CALM) that allow learning of temporal sequences are introduced in this paper. We embedded an associative learning mechanism which allows to look into the past when classifying present stimuli. We have built in the Euclidean metrics instead of the weighted sum found in the original learning rule. This improvement allows better discrimination in case of learning low dimensional patterns in the temporal sequences. Results were obtained from testing the enhanced module on simple artificial data. These experiments promise applicability of the enhanced module in a real problem domain. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • RTD-based compact programmable gates

    Publication Year: 2004 , Page(s): 2637 - 2640 vol.4
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (606 KB) |  | HTML iconHTML  

    This work presents novel and extremely compact implementations of programmable gates on the basis of the multi-threshold threshold gate concept. The circuit consists of resonant tunnelling diodes (RTDs) and heterostructure field effect transistors (HFETs) and its operating principle is based on the controlled quenching of clocked series-connected RTDs. The proposed generic circuit topology is presented and the methodology to design specific programmable gates is introduced. A number of programmable gates are shown and their operation is validated. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A SVCA model for the competition on artificial time series (CATS) benchmark

    Publication Year: 2004 , Page(s): 2777 - 2782 vol.4
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (733 KB) |  | HTML iconHTML  

    This paper predicts the 100 missing values in CATS Benchmark. The SVCA model is an autoregressive model in which the coefficients vary smoothly with time. The model is fitted to the first differences of the data by minimising the residual sum of squared, subject certain restrictions that enable the gaps left by the missing observations to be bridged. The path of each time-varying coefficient is described by a combination of a sine and cosine function. The latter are specified via their amplitudes, phases and periods. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Neural neutron/gamma discrimination in organic scintillators for fusion applications

    Publication Year: 2004 , Page(s): 2931 - 2936 vol.4
    Cited by:  Papers (3)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (684 KB) |  | HTML iconHTML  

    This work deals with the discrimination of neutrons and gamma rays on the basis of their different pulse shapes in scintillator detectors; this technique is widely employed in nuclear fusion applications. After a thorough phase of data analysis, a multi layer perceptron (MLP) is trained with the aim of processing the shape of light pulses produced by these ionizing particles in an organic liquid scintillator and digitally acquired. Moreover, fast-superimposed events (called pile-ups) are detected and a further MLP is trained to analyze them and recover the original superimposed events. Satisfactory experimental results were obtained at the Frascati Tokamak Upgrade, ENEA-Frascati, Italy. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Support vector classifiers via gradient systems with discontinuous righthand sides

    Publication Year: 2004 , Page(s): 2997 - 3002 vol.4
    Cited by:  Papers (2)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (711 KB) |  | HTML iconHTML  

    This paper implements support vector machines (SVM) for the discrimination of nonseparable classes using gradient systems with discontinuous righthand sides. The gradient systems are obtained from an exact penalty method applied to the constrained quadratic optimization problems. Global convergence to the solution of the corresponding constrained problems is shown to be independent of the penalty parameters and of the regularization parameter of the SVM. View full abstract»

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

    Publication Year: 2004 , Page(s): 2825 - 2830 vol.4
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (731 KB) |  | HTML iconHTML  

    It is common to have both observed and missing values in data. This paper concentrates on the case where a value can be somewhere between those two ends, partially observed and partially missing. To achieve that, a method of using evidence nodes in a Bayesian network is studied. Different ways of handling inaccuracies are discussed in examples and the proposed approach is justified in the experiments with real image data. Also, a justification is given for the standard preprocessing step of adding a tiny amount of noise to the data, when a continuous valued model is used for discrete-valued data. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • The support vector machine learning using the second order cone programming

    Publication Year: 2004 , Page(s): 2991 - 2996 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (643 KB) |  | HTML iconHTML  

    We propose a data dependent learning method for the support vector machine. This method is based on the technique of second order cone programming. We reformulate the SVM quadratic problem into the second order cone problem. The proposed method requires decomposing the kernel matrix of SVM optimization problem. In this paper we apply Cholesky decomposition method. Since the kernel matrix is positive semi definite, some columns of the decomposed matrix diminish. The performance of the proposed method depends on the reduction of dimensionality of the decomposed matrix. Computational results show that when the columns of decomposed matrix are small enough, the proposed method is much faster than the quadratic programming solver LOQO. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Kernel-based canonical coordinate decomposition of two-channel nonlinear maps

    Publication Year: 2004 , Page(s): 3019 - 3024 vol.4
    Cited by:  Papers (4)  |  Patents (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (720 KB) |  | HTML iconHTML  

    A kernel-based formulation for decomposing nonlinear maps of two data channels into their canonical coordinates is derived. Each data channel is implicitly mapped to a high dimensional feature space defined by a nonlinear kernel. The canonical coordinates of the nonlinear maps are then found by transforming the kernel maps with the eigenvector matrices of a coupled asymmetric generalized eigenvalue problem. This generalized eigenvalue problem is constructed in the explicit space of kernel maps. The measures of linear dependence and coherence between the nonlinear maps of the channels are also presented. These measures may be determined in the kernel domain, without explicit computation of the nonlinear mappings. A numerical example is also presented. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Interdisciplinary research and speech rhythm

    Publication Year: 2004 , Page(s): 2729 - 2733 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (649 KB) |  | HTML iconHTML  

    The new definition of speech rhythm is given. This definition is based on the principle of segmentation: voiced segment - unvoiced segment. The rhythm is defined as sequence of these segments' durations. Set of normalised durations is the point set on interval [0, 1]. Owing to established in paper fact that the rhythm evolution is described as the logistic mapping on this set, in the offered paper the dynamic regimes of the rhythm in both normal speech and stuttering speech are described. It is shown that such definition of the rhythm is physiologically plausible in contrast to earlier reported models and allows to explain the nature of stuttering which did not find explanation up to now. It turned out that the nets offered by van der Maas have exactly same route to chaos in dependence on the changing control parameter or parameter of "learning" in the net as the logistic map with the same control parameter has. Moreover, within the framework of the offered mathematical model the causes, which change these rhythm regimes, are established. It allowed to apply in clinic optimal course of treatment for each stutterer individually (Skljarov, O.P. and Poroshin, A.N., Jan. 2003). View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Artificial neural networks, where do we go next?

    Publication Year: 2004 , Page(s): 2989 - 2990 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (438 KB)  

    Summary form only given. The possibilities of going to the next level after neural networks by creating new intelligent complex systems are discussed in This work. To create such complex systems, there are a number of important problems that must be solved. The problems presented are as follows: (a) scaling; (b) the degree of biological accuracy; (c) how to do system integration; and (d) hardware acceleration. Some examples of research in each area were also given in This work. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Histogram coding for recognition of contours presented by Bezier curves

    Publication Year: 2004 , Page(s): 2559 - 2563 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (708 KB) |  | HTML iconHTML  

    In the pattern recognition area, one of the most important tasks is the ability of a neural network to classify objects regardless of affine transformations. Contoured objects can be described with Bezier curves and the description is affine transformation invariant. Direct use of the curves for a neural network input isn't applicable because it's possible that descriptions of the same objects consist of different number of Bezier curves. We propose histogram coding, decomposing a list of Bezier curves, which can be used as an input for a neural network. Experiments show that proposed coding gives good results to solve formulated problem. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Transformation-invariant representation and NMF

    Publication Year: 2004 , Page(s): 2535 - 2539 vol.4
    Cited by:  Papers (5)  |  Patents (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (960 KB) |  | HTML iconHTML  

    Non-negative matrix factorization (NMF) is a method for the decomposition of multivariate data into strictly positive activations and basis vectors. Here, instead of using unstructured data vectors, we assume that something is known in advance about the type of transformations that either the input data or the basis vectors may undergo. This would be the case e.g. if we assume input vectors that are translationally shifted versions of each other, but it applies to any other transformations as well. The key idea is that we factorize the data into activations and basis vectors modulo the transformations. We show that this can be done by extending NMF in a natural way. The gained factorization thus provides a transformation-invariant and compact encoding that is optimal for the given transformation constraints. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Visual comparison of performance for different activation functions in MLP networks

    Publication Year: 2004 , Page(s): 2947 - 2952 vol.4
    Cited by:  Papers (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (702 KB)  

    Multi layer perceptron networks have been successful in many applications, yet there are many unsolved problems in the theory. Commonly, sigmoidal activation functions have been used, giving good results. The backpropagation algorithm might work with any other activation function on one condition though - it has to have a differential. We investigate some possible activation functions and compare the results they give on some sample data sets. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Structural hierarchies, theta rhythm, hippocampal function

    Publication Year: 2004 , Page(s): 3047 - 3051 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (657 KB) |  | HTML iconHTML  

    The connection between different structures being at different hierarchical level of the cortico-hippocampal formation and their functional role is discussed. At least three different functions, code generation, mood regulation and navigation is being integrated into a coherent conceptual framework. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Nonlinear oscillation models for the spike sorting of single units recorded extracellularly

    Publication Year: 2004 , Page(s): 3029 - 3033 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (653 KB) |  | HTML iconHTML  

    The present study is devoted to the problem of automatic sorting of extracellularly recorded action potentials of neurons. The classification of spike waveform is considered as a pattern recognition problem of segments of signal that corresponds to the appearance of spikes. Nonlinear oscillating model with perturbation is used to describe the waveforms of spikes. It allows characterizing the signal distortions in both amplitude and phase. The spikes generated by one neuron assumed to be described by the same equation and should be recognized as one class. The problem of spike recognition is reduced to the separation of mixture of normal distributions in the transformed feature space. An unsupervised iteration-learning algorithm that estimates the number of classes and their centers is developed. It scans the learning set in order to evaluate spikes trajectories in phase space with maximal probability density in their neighborhood. To estimate the trajectories the integral operators with piece-wise polynomial kernels were used that provides computational efficiency. The new algorithm was tested on simulated and real data sets. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Option pricing and trading with artificial neural networks and advanced parametric models with implied parameters

    Publication Year: 2004 , Page(s): 2741 - 2746 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (744 KB) |  | HTML iconHTML  

    We combine parametric models and feedforward artificial neural networks to price and trade European S&P500 Index options. Artificial neural networks are optimized on a hybrid target function consisted by the standardized residual term between the actual market price and the option estimate of a certain parametric model. Parametric models include: (i) the Black and Scholes model that assumes a geometric Brownian motion process (GBM); (ii) the Corrado and Su that additionally allows for excess skewness and kurtosis via a Gram-Charlier series expansion; (iii) analytic models that extend the GBM by incorporating multiple sources of Poisson distributed jumps; and (vi) stochastic volatility and jump models. Daily average implied parameters of these models are estimated with options transaction data via an unconstraint process optimized by the non-linear least squares Levenberg-Marquardt algorithm. This structural average implied parameters are used to validate the out-of sample pricing and trading (with transaction costs) ability of all models developed. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Acoustic model combination for recognition of speech in multiple languages using support vector machines

    Publication Year: 2004 , Page(s): 3065 - 3069 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (625 KB) |  | HTML iconHTML  

    We study the performance of support vector machine based classifiers in acoustic model combination for recognition of context dependent sub word units of speech in multiple languages. In acoustic model combination, the data for similar sub word units across languages are shared to train acoustic models for multilingual speech. Sharing of data across languages leads to an increase in the number of training examples for a subword unit common to the languages. It may also lead to increase in the variability of the data for a subword unit. In This work, we study the effect of data sharing on the classification accuracy and complexity of acoustic models built using support vector machines. We compare the performance of multilingual acoustic models with that of monolingual acoustic models in the recognition of a large number of consonant-vowel units in the broadcast news corpus of three Indian languages. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Expert systems and artificial neural networks applied to stellar optical spectroscopy: a comparative analysis

    Publication Year: 2004 , Page(s): 2705 - 2710 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (748 KB) |  | HTML iconHTML  

    This work presents a comparative study of two computational techniques - expert systems and artificial neural networks - applied to a specific field of astrophysics, the classification of the optical spectra of stars. We present a description of various expert systems and neural networks models, and the comparison of the results obtained by each technique individually and by a combination of both. We do not only intend to analyse the efficiency of these two approaches in the classification of stellar spectra; our final objective is the integration of several techniques in a unique hybrid system. This system will be capable of applying the most appropriate classification method to each spectrum, which widely opens the research in the field of automatic spectral classification. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Prototype optoelectronic Hamming neural network

    Publication Year: 2004 , Page(s): 2659 - 2663 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (663 KB) |  | HTML iconHTML  

    We describe the hardware implementation of a Hamming classifier using an optoelectronic architecture. It is composed of two layers, the first layer is an optoelectronic matrix-vector multiplier based on the optical broadcast architecture; it is a novel architecture composed of a set of electronic neurons that receive the input sequentially by means of an optical broadcast interconnection. The second layer is an electronic winner take all circuit. The main characteristic of the system is that it is readily scalable in speed and size to large numbers of pixel neurons. We will describe the optoelectronic architecture, the hardware implementation of a prototype and evaluation of its performance characteristics. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Sparse coding and NMF

    Publication Year: 2004 , Page(s): 2529 - 2533 vol.4
    Cited by:  Papers (26)  |  Patents (1)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (906 KB) |  | HTML iconHTML  

    Non-negative matrix factorization (NMF) is a very efficient parameter-free method for decomposing multivariate data into strictly positive activations and basis vectors. However, the method is not suited for overcomplete representations, where usually sparse coding paradigms apply. We show how to merge the concepts of non-negative factorization with sparsity conditions. The result is a multiplicative algorithm that is comparable in efficiency to standard NMF, but that can be used to gain sensible solutions in the overcomplete cases. This is of interest e.g. for the case of learning and modeling of arrays of receptive fields arranged in a visual processing map, where an overcomplete representation is unavoidable. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Stochastic dynamics and partial synchronization of stimulus-driven neural activity

    Publication Year: 2004 , Page(s): 3035 - 3040 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (768 KB) |  | HTML iconHTML  

    We study a dynamical behaviour of neural population driven by stimulus. This dynamical response is considered in relation to the population coding. The hypothesis that neuronal code at some stages of information processing in the brain is based on synchronisation of neural activity is under intensive discussions. The dynamical regime of partial synchronisation is important and very useful for modelling of neural activity and we have found that the input driven neural assembly can demonstrate a dynamical regime of partial synchronisation. It is interesting to note that population dynamics has a stochastic nature and repetition of the same stimulus causes a synchronous activity of different sub-populations. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Recognizing objects in non-controlled backgrounds by an appearance two-step approach

    Publication Year: 2004 , Page(s): 2565 - 2570 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (673 KB) |  | HTML iconHTML  

    This work presents a method for identifying real three-dimensional objects in non-controlled backgrounds using independent component analysis to eliminate redundant image information present in each object image. The proposed method is a two-step process that allows a coarse color-based detection and an exact localization using shape information. The paper describes an efficient implementation, making this approach suitable for real-time applications. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Comparative performance of neural networks and maximum likelihood for supervised classification of agricultural crops: single date and temporal analysis

    Publication Year: 2004 , Page(s): 2959 - 2964 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (722 KB) |  | HTML iconHTML  

    Maximum likelihood, backpropagation and radial basis neural networks were applied in the supervised classification of agricultural crops. Ten ETM+t/Landsat rectified images in bands 3, 4, 5 and NDVI were used as input data for the classification. The NDVI input was used as an indicator for changes in the leaf area index and, by correlation, the phenological cycle. Agriculture in the study area makes the spectral characterization of dry season crops troublesome since irrigation possibilities give the planting date flexibility, while the phenological stages in training polygons are rarely representative of the whole image. Kappa statistics showed that temporal classification, which analyses a pixel in continuum, improved the discrimination in comparison to a single spectral date at a significant level (p < 0.05) in many dates. The neural network models (multilayer perceptron and radial basis functions) had a very similar performance that surpassed the maximum likelihood method. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • TLS linear prediction with optimum root selection for resolving closely space sinusoids

    Publication Year: 2004 , Page(s): 2699 - 2703 vol.4
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (600 KB) |  | HTML iconHTML  

    Total least square linear prediction has been successfully applied to frequency estimation for closely spaced sinusoids. In low signal to noise ratio, the resolving ability of TLS is degraded and extraneous roots of the predictor are close to unit circle. Hence the performance of total least square is severely degraded in low SNR. In this paper, a generalized total least squares method with a new root selection criterion, which is based on the envelope of the signal spectrum, is presented. An optimum procedure is introduced to provide a TLS solution that can perform closer to Cramer-Rao bound, particularly in low SNR. View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Symbol grounding transfer with hybrid self-organizing/supervised neural networks

    Publication Year: 2004 , Page(s): 2865 - 2869 vol.4
    Cited by:  Papers (2)
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (687 KB) |  | HTML iconHTML  

    This paper reports new simulations on an extended neural network model for the transfer of symbol grounding. It uses a hybrid and modular connectionist model, consisting of an unsupervised, self-organizing map for stimulus classification and a supervised network for category acquisition and naming. The model is based on a psychologically-plausible view of symbolic communication, where unsupervised concept formation precedes the supervised acquisition of category names. The simulation results demonstrate that grounding is transferred from symbols denoting object properties to newly acquired symbols denoting the object as a whole. The implications for cognitive models integrating neural networks and multi-agent systems are discussed. View full abstract»

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