Simulação do comportamento estocástico do algoritmo KLMS com diferentes kernels

The kernel least-mean-square (KLMS) algorithm is a popular algorithmin nonlinear adaptive filtering due to its simplicity androbustness. In kernel adaptive filtering, the statistics of the inputto the linear filter depends on the kernel and its parameters. Moreover,practical implementations on systems estimation require afinite non-linearity model order. In order to obtain finite ordermodels, many kernelized adaptive filters use a dictionary of kernelfunctions. Dictionary size also depends on the kernel and itsparameters. Therefore, KLMS may have different performanceson the estimation of a nonlinear system, the time of convergence,and the accuracy using a different kernel. In order to analyze theperformance of KLMS with different kernels, this paper proposesthe use of the Monte Carlo simulation of both steady-state and thetransient behavior of the KLMS algorithm using different types ofkernel functions and Gaussian inputs.

Dynamic evolution of sentiments in Never Let Me Go: Insights from multifractal theory and its implications for literary analysis

The moods, feelings, and attitudes represented in a novel will resonate in the reader by activating similar sentiments. It is generally accepted that sentiment analysis can capture aspects of such moods, feelings, and attitudes and can be used to summarize a novel’s plot in a story arc. With the availability of a number of algorithms to automatically extract sentiment-based story arcs, new approaches for their utilization becomes pertinent. We propose to use nonlinear adaptive filtering and fractal analysis in order to analyze the narrative coherence and dynamic evolution of a novel. Using Never Let Me Go by Kazuo Ishiguro, the winner of the 2017 Nobel Prize for Literature as an illustrative example, we show that: (1) nonlinear adaptive filtering can extract a story arc that reflects the tragic trend of the novel; (2) the story arc displays persistent dynamics as measured by the Hurst exponent at short and medium timescales; (3) the plot’s dynamic evolution is reflected in the time-varying Hurst exponent. We argue that these findings are indicative of the potential that multifractal theory has for computational narratology and large-scale literary analysis. Specifically that the global Hurst exponent of a story arc is an index of narrative coherence that can identify bland, incoherent, and coherent narratives on a continuous scale. And, further, that the local time-varying Hurst exponent captures variation of a novel’s plot such that the extrema have specific narratological interpretations.

A Kernel Recursive Maximum Versoria-Like Criterion Algorithm for Nonlinear Channel Equalization

In this paper, a kernel recursive maximum Versoria-like criterion (KRMVLC) algorithm has been constructed, derived, and analyzed within the framework of nonlinear adaptive filtering (AF), which considers the benefits of logarithmic second-order errors and the symmetry maximum-Versoria criterion (MVC) lying in reproducing the kernel Hilbert space (RKHS). In the devised KRMVLC, the Versoria approach aims to resist the impulse noise. The proposed KRMVLC algorithm was carefully derived for taking the nonlinear channel equalization (NCE) under different non-Gaussian interferences. The achieved results verify that the KRMVLC is robust against non-Gaussian interferences and performs better than those of the popular kernel AF algorithms, like the kernel least-mean-square (KLMS), kernel least-mixed-mean-square (KLMMN), and Kernel maximum Versoria criterion (KMVC).

Kernel Affine Projection P-norm (KAPP) Filtering under Alpha Stable Distribution Noise Environment

Aiming at improving the performance of the nonlinear adaptive filtering under the alpha-stable distribution noise environment, Kernel Affine Projection P-norm (KAPP) algorithm based on the minimum dispersion coefficient criterion and the affine projection is deduced. The accuracy of the gradient estimation is enhanced by using the input signals and the error signals at multiple times. The simulation results on Mackey–Glass chaotic time series prediction show that the KAPP algorithm has faster convergence speed, better steady-state performance and stronger robustness under the Gaussian noise and stable distributed noise environment.