Showing papers in "Neurocomputing in 2005"

TL;DR: Experimental results on ORL and a subset of FERET face databases show that (2D)^2PCA achieves the same or even higher recognition accuracy than 2DPCA, while the former needs a much reduced coefficient set for image representation than the latter.

TL;DR: It is demonstrated on benchmark datasets that the CMA-ES improves the results achieved by grid search already when applied to few hyperparameters and that tuning of the scaling and the rotation of Gaussian kernel can lead to better results in comparison to standard Gaussian kernels with a single bandwidth parameter.

TL;DR: The simulation results show that the ELM equalizer significantly outperforms other neural network equalizers such as the complex minimal resource allocation network (CMRAN), complex radial basis function (CRBF) network and complex backpropagation (CBP) equalizers.

TL;DR: In the present paper, models of the water level-discharge relationship are built with an artificial neural network (ANN) and an M5 model tree-based models that are superior in accuracy than the traditional model.

TL;DR: The results show that the proposed SAPSO- based ANN has a better ability to escape from a local optimum and is more effective than the conventional PSO-based ANN.

TL;DR: The use of the data-driven empirical mode decomposition (EMD) method to study neuronal activity in visual cortical area V4 of macaque monkeys performing a visual spatial attention task found that local field potentials were resolved by the EMD into the sum of a set of intrinsic components with different degrees of oscillatory content.

TL;DR: The natural gradient method for neural networks is extended to the case where the weight vectors are constrained to the Stiefel manifold and a simpler updating rule is developed and one parameter family of its generalizations is developed.

TL;DR: Aw-SpPCA can adaptively compute the contributions of each part of the human face and then endows them to a classification task in order to enhance the robustness to both expression and illumination variations.

TL;DR: Cortical neurons extracted from the developing rat central nervous system and put in culture show spontaneous activity with a typical electrophysiological pattern ranging from stochastic spiking to synchronized bursting, and significant changes have been revealed in the firing dynamics at different stages of the developmental process.

TL;DR: Simulation results in six problems of the PROBEN1 benchmark collection show that the globally convergent modification of the Rprop algorithm exhibits improved learning speed, and compares favorably against the original Rprop and the Improved Rprop, a recently proposed Rrpop modification.

TL;DR: The context model is combined with the neural gas vector quantizer to obtain merging neural gas (MNG) for temporal data and theoretic results on the representation capabilities and the MSOM training dynamic are presented.

TL;DR: A sufficient condition for the existence, uniqueness and global asymptotic stability of the equilibrium point for bidirectional associative memory (BAM) neural networks with fixed time delays is presented.

TL;DR: This paper explores the possibility of providing robots with an 'inner world' based on internal simulation of perception rather than an explicit representational world model, and finds that a trained robot is in some cases actually able to move blindly in a simple environment for hundreds of time steps, successfully handling several multi-step turns.

TL;DR: This paper introduces a model that uses a different architecture compared to the traditional neural network, to capture and forecast nonlinear processes, and shows that this approach performs well when compared with traditional models and established research.

TL;DR: Discrete-time versions of the continuous-time Cohen-Grossberg neural networks are formulated and studied and sufficient conditions are obtained to ensure the global exponential stability of the discrete-time systems of CGNNs with and without delays based on Lyapunov methods.

TL;DR: A class of simple chaotic Hopfield neural networks is presented, and a bifurcation from transient chaos to chaos is discussed.

TL;DR: It is demonstrated how a variational approximation scheme enables effective inference of key parameters in probabilisitic signal models which employ the Student-t distribution.

TL;DR: A simple and fast procedure that can tear or cut 'circular' manifolds, i.e. break their essential loops, in order to make their embedding in a low-dimensional space easier.

TL;DR: The proposed feature selection scheme is able to effectively identify the salience features and is compared with the related study through applying to different classification problems in which the number of features ranged from less than 10 to over 12,600.

TL;DR: A novel subspace method, named supervised kernel locality preserving projections (SKLPP), is proposed for face recognition, in which geometric relations are preserved according to prior class-label information and complex nonlinear variations of real face images are represented by nonlinear kernel mapping.

TL;DR: It is shown that multichannel time encoding using filter banks and integrate-and-fire neurons provides an invertible representation of information, i.e., a sensory stimulus can be recovered from its multidimensional spike train representation loss-free.

TL;DR: This work demonstrates that a system based on K-local hyperplane outperforms the methods proposed in the literature based on global representation of a protein pair in predicting protein-protein interaction.

TL;DR: The use of geometrical methods to tackle the non-negative independent component analysis (non-negative ICA) problem, without assuming the reader has an existing background in differential geometry, is explored.

TL;DR: An expectation-maximization (EM) algorithm that yields topology preserving maps of data based on probabilistic mixture models that allows principled handling of missing data and learning of mixtures of SOMs.

TL;DR: A different approach based on random subspace ensembles of SVMs: a set of base learners is trained and aggregated using subsets of features randomly drawn from the available DNA microarray data shows the effectiveness of the proposed approach.

TL;DR: This work proposes a general regression framework, based on restriction of the search space to a subspace and a particular choice of basis vectors in feature space, which allows to accommodate kernel Partial Least Squares and kernel Canonical Correlation analysis for regression with a sparse representation, which makes them applicable to large data sets, with little loss in accuracy.

TL;DR: This study on the inverse problem of SVMs is motivated by designing a heuristic algorithm for generating decision trees with high generalization capability.

TL;DR: Experimental results obtained on the SUBCORPUS-100 MCYT signature database show that the machine expert outperforms the state-of-the-art works both for random and skilled forgeries.

TL;DR: Insight is provided into the organization and dynamics of recurrent online training algorithms by comparing real time recurrent learning (RTRL) with a new continuous-time online algorithm, which is derived in the spirit of a recent approach introduced by Atiya and Parlos (APRL), which leads to non-gradient search directions.

TL;DR: In this paper, hyperchaos is demonstrated in a classical Hopfield-type neural network with four neurons for some weight matrices with unknown values for each neuron.