Showing papers in "Neural Networks in 2012"

TL;DR: A publicly available traffic sign dataset with more than 50,000 images of German road signs in 43 classes is presented, and Convolutional neural networks showed particularly high classification accuracies in the competition, and the CNNs outperformed the human test persons.

TL;DR: This work uses a fast, fully parameterizable GPU implementation of a Deep Neural Network (DNN) that does not require careful design of pre-wired feature extractors, which are rather learned in a supervised way.

TL;DR: Analytical examples are used to show that a widely used criterion for the ESP, the spectral radius of the weight matrix being smaller than unity, is not sufficient to satisfy the echo state property.

TL;DR: Based on the stability analysis, the critical values of the fractional order for which Hopf bifurcations may occur are identified and Simulation results are presented to illustrate the theoretical findings and to show potential routes towards the onset of chaotic behavior when the fractions of the system increases.

TL;DR: The paper shows that current limitations of optical flow computation can be overcome by using event-based visual acquisition, where high data sparseness and high temporal resolution permit the computation of Optical flow with micro-second accuracy and at very low computational cost.

TL;DR: A novel Laplacian Twin Support Vector Machine (called Lap-TSVM) is proposed for the semi-supervised classification problem, which can exploit the geometry information of the marginal distribution embedded in unlabeled data to construct a more reasonable classifier and be a useful extension of TSVM.

TL;DR: A general class of memristor-based recurrent neural networks with time-varying delays with exponential convergence and conditions on the nondivergence and global attractivity are established by using local inhibition.

TL;DR: Results show that a simple biologically inspired unsupervised learning scheme is capable of generating selectivity to complex meaningful events on the basis of relatively little sensory experience.

TL;DR: In this paper, the authors present a survey of kernel-on-graphs (kernels on graphs) and two related similarity matrices, which they refer to as kernels on graph.

TL;DR: In this article, the existence and α -exponential stability of the equilibrium point of fractional-order neural networks is investigated. But the authors only considered fractional order neural networks and did not consider fractional chaotic networks.

TL;DR: A one-layer recurrent neural network is proposed for solving pseudoconvex optimization problems subject to linear equality and bound constraints and is capable of solving constrained fractional programming problems as a special case.

TL;DR: A new Twin Support Vector Machine with Universum (called U-TSVM), which can utilize Universum data to improve the classification performance of TSVM by using two Hinge Loss functions.

TL;DR: The method of stochastic stability is applied to derive the thermodynamics of the model, and it is implied that simulating the dynamics of a Hopfield network can be accomplished by a hybrid Boltzmann Machine, requiring the update of N+P neurons but the storage of only NP synapses.

TL;DR: All of the model transformations, cross-terms bounding techniques and free additional matrix variables are avoided in the derivation, so the results obtained have less conservatism and simpler formulations than the existing ones.

TL;DR: A comparative analysis of the basic ELMs and support vector machines from two viewpoints that are different from previous works: one is the Vapnik-Chervonenkis (VC) dimension, and the other is their performance under different training sample sizes.

TL;DR: An extension of the recently introduced Generalized Matrix Learning Vector Quantization algorithm to matrices of limited rank corresponding to low-dimensional representations of the data to incorporate prior knowledge of the intrinsic dimension and to reduce the number of adaptive parameters efficiently.

TL;DR: A new delay-dependent exponential stability condition is proposed, such that for all admissible delay bounds, the resulting estimation error system is mean-square exponentially stable with a prescribed noise attenuation level in the H(∞) sense.

TL;DR: This paper provides a uniformly ultimately boundedness (UUB) result for the direct HDP learning controller under mild and intuitive conditions and shows that the estimation errors of the learning parameters or the weights in the action and critic networks remain UUB.

TL;DR: This paper proves that the variance of gradient estimates in the PGPE (policy gradients with parameter-based exploration) method is smaller than that of the classical REINFORCE method under a mild assumption, and derives the optimal baseline for PGPE, which contributes to further reducing the variance.

TL;DR: The proposed criterion for exponential synchronization generalizes and improves those reported recently in the literature and is presented as an illustrative example to show the feasibility and effectiveness of the proposed scheme.

TL;DR: The existence of a unique equilibrium point of a generalized Cohen-Grossberg BAM neural networks of neutral type delays is discussed by means of the Homeomorphism theory and inequality technique and the global asymptotic stability of the equilibrium solution to the above neural networks is studied.

TL;DR: A coordinate descent margin based twin vector machine (CDMTSVM) compared with the original TWSVM is presented and a novel coordinate descent method is proposed for the dual problems which leads to very fast training.

TL;DR: A new ϵ-optimal control algorithm based on the iterative ADP approach is proposed that makes the performance index function iteratively converge to the greatest lower bound of all performance indices within an error ϵ in finite time.

TL;DR: This paper addresses the stability problem of a class of delayed neural networks with time-varying impulses by investigating both the stabilizing and destabilizing impulses considered simultaneously based on the comparison principle.

TL;DR: It is demonstrated that training recurrent networks engineered for specific tasks can produce better results than single-layer networks and LSTM-g has the potential to both improve the performance and broaden the applicability of spatially and temporally local gradient-based training algorithms for recurrent neural networks.

TL;DR: A novel nonparallel plane classifier, called Weighted Twin Support Vector Machines with Local Information (WLTSVM), which mines as much underlying similarity information within samples as possible and has its additional advantages.

TL;DR: A stability analysis is conducted by exploiting the stability theory of Lyapunov functionals and the theory of Homomorphic mapping to derive some easily verifiable sufficient conditions for existence, uniqueness and global asymptotic stability of the equilibrium point for the class of neural networks with discrete constant time delays under parameter uncertainties.

TL;DR: A very interesting fact is revealed that the synchronization of neural networks with reaction-diffusions is more easily realized than those of Neural networks without reaction-Diffusions.

TL;DR: This study proposes a novel method of spectral deconvolution based on Bayesian estimation with the exchange Monte Carlo method, which is an application of the integral approximation of stochastic complexity and the exchangeMonte Carlo method.

TL;DR: New theoretical results are presented on the global robust exponential stability of interval fuzzy Cohen-Grossberg networks with piecewise constant argument on the basis of the comparison principle.