Yatong Zhou

Analysis of the Distance Between Two Classes for Tuning SVM Hyperparameters

An important step in the construction of a support vector machine (SVM) is to select optimal hyperparameters. This paper proposes a novel method for tuning the hyperparameters by maximizing the distance between two classes (DBTC) in the feature space. With a normalized kernel function, we find that DBTC can be used as a class separability criterion since the between-class separation and the within-class data distribution are implicitly taken into account. Employing DBTC as an objective function, we develop a gradient-based algorithm to search the optimal kernel parameter. On the basis of the geometric analysis and simulation results, we find that the optimal algorithm and the initialization problem become very simple. Experimental results on the synthetic and real-world data show that the proposed method consistently outperforms other existing hyperparameter tuning methods.

Spike-Like Blending Noise Attenuation Using Structural Low-Rank Decomposition

Spikelike noise is a common type of random noise existing in many geoscience and remote sensing data sets. The attenuation of spike-like noise has become extremely important recently, because it is the main bottleneck when processing the simultaneous source data that are generated from the modern seismic acquisition. In this letter, we propose a novel low-rank decomposition algorithm that is effective in rejecting the spike-like noise in the seismic data set. The specialty of the low-rank decomposition algorithm is that it is applied along the morphological direction of the seismic data sets with a prior knowledge of the morphology of the seismic data, which we call local slope. The seismic data are of much lower rank along the morphological direction than along the space direction. The morphology of the seismic data (local slope) is obtained via a robust plane-wave destruction method. We use two simulated field data examples to illustrate the algorithm workflow and its effective performance.

Ground-Roll Noise Attenuation Using a Simple and Effective Approach Based on Local Band-Limited Orthogonalization

Bandpass filtering is a common way to estimate ground-roll noise on land seismic data, because of the relatively low-frequency content of ground roll. However, there is usually a frequency overlap between ground roll and the desired seismic reflections that prevents bandpass filtering alone from effectively removing ground roll without also harming the desired reflections. We apply a bandpass filter with a relatively high upper bound to provide an initial imperfect separation of ground roll and reflection signal. We then apply a technique called “local orthogonalization” to improve the separation. The procedure is easily implemented, since it involves only bandpass filtering and a regularized division of the initial signal and noise estimates. We demonstrate the effectiveness of the method on an open-source set of field data.

Letters: Nonlinear noise reduction of chaotic time series based on multidimensional recurrent LS-SVM

In order to resolve the noise reduction in chaotic time series, a novel method based on multidimensional recurrent least squares support vector machine (MDRLS-SVM) is proposed in this paper. Considering the evolvement feature of the chaotic system, we utilize the recurrent version of least squares support vector machines (LS-SVM) to manipulate the iterative problem. From the high-dimensional phase space point of view, the function approximation in the high-dimensional embedding phase space is carried out and the noise reduction achieved simultaneously based on the reconstructed embedding phase theory. We show by means of simulation of Ikeda map that the proposed method is able to provide accurate results in noise reduction of chaotic system.

A One-Step Network Traffic Prediction

In the information society today computer networks are an indispensable part of peoples life. Network traffic prediction is important to network planning, performance evaluation and network management directly. A variety of machine learning models such as artificial neural networks (ANN) and support vector machine (SVM) have been applied in traffic prediction. In this paper, a novel network traffic one-step-ahead prediction technique is proposed based on a state-of-the-art learning model called minimax probability machine (MPM). The predictive performance is tested on traffic data of Ethernet, experimental results show that the predictions of MPM match the actual traffics accurately and the proposed methods can increases the computational efficiency. Furthermore, we compare the MPM-based prediction technique with the SVM-based techniques. The results show that the predictive performance of MPM is competitive with SVM.

Constant Modulus Based Blind Channel Estimation for OFDM Systems

A novel blind channel estimation scheme is proposed for OFDM systems employing PSK modulation. This scheme minimizes the number of possible channels by exploiting the constant modulus property, chooses a best fit over the possible channels by exploiting the finite alphabet property of information signals, and achieves competitive performance with low computational complexity. Results comparing the new scheme with the finite-alphabet based channel estimation are presented.

A novel Least Squares Support Vector Machine kernel for approximation

The support vector machine (SVM) is receiving considerable attention for its superior ability to solve nonlinear classification, function estimation and density estimation. Least squares support vector machines (LS-SVM) are re-formulations to the standard SVMs. Motivated by the theory of multi-scale representations of signals and wavelet transforms, this paper presents a way for building a wavelet-based reproducing kernel Hilbert spaces (RKHS) and its associate scaling kernel for least squares support vector machines (LS-SVM). The RKHS built is a multiresolution scale subspace, and the scaling kernel is constructed by using a scaling function with its different dilations and translations. Compared to the traditional kernels, approximation results illustrate that the LS-SVM with scaling kernel enjoys two advantages: (1) it can approximate arbitrary signal and owns better approximation performance; (2) it can implement multi-scale approximation.

Predicting nonstationary time series with multi-scale gaussian processes model

The Gaussian processes (GP) model has been successfully applied to the prediction of nonstationary time series. Due to the models covariance function containing an undetermined hyperparameters, to find its maximum likelihood values one usually suffers from either susceptibility to initial conditions or large computational cost. To overcome the pitfalls mentioned above, at the same time to acquire better prediction performance, a novel multi-scale Gaussian processes (MGP) model is proposed in this paper. In the MGP model, the covariance function is constructed by a scaling function with its different dilations and translations, ensuring that the optimal value of the hyperparameter is easy to determine. Although some more time is spent on the calculation of covariance function, MGP takes much less time to determine hyperparameter. Therefore, the total training time of MGP is competitive to GP. Experiments demonstrate the prediction performance of MGP is better than GP. Moreover, the experiments also show that the performance of MGP and support vector machine (SVM) is comparable. They give better performance compared to the radial basis function (RBF) networks.

Support Vector Machine Based Multiresolution Signal Approximation

Multiresolution signal approximation (MSA) provides a simple hierarchical approximation of the signals. And support vector machine (SVM) has been introduced as a novel tool for solving approximation problems. Based on the fact that scale subspaces onto which MSA projects the signals are reproducing kernel Hilbert spaces (RKHS), we integrate the approximation criterion of SVM into MSA and then an SVM based MSA (S-MSA) algorithm is proposed. Experiments exhibit that S-MSA owns better approximation accuracy and smoothness than MSA. Furthermore, quantitative comparison with MSA illustrates the robustness of S-MSA when noises are present

Music style classification with a novel bayesian model

Music style classification by mean of computers is very useful to music indexing, content-based music retrieval and other multimedia applications. This paper presents a new method for music style classification with a novel Bayesian-inference-based decision tree (BDT) model. A database of total 320 music staffs collected from CDs and the Internet is used for the experiment. For classification three features including the number of sharp octave (NSO), the number of simple meters (NSM), and the music playing speed (MPS) are extracted. Following that, acomparative evaluation between BDT and traditional decision tree (DT) model is carried out on the database. The results show that the classification accuracy rate of BDT far superior to existing DT model.