Gingivitis Classification via Wavelet Entropy and Support Vector Machine

Wavelet neural networks (WNN) and support vector machine (SVM) are two advanced methods which are fit for classification. A comparative analysis of the two methods was conducted based on Chinese firms. The results show WNN has good classification effect. Wavelet decomposition has been demonstrated to be an effective tool for recognizing the firms' feature. Also the study applied SVM to the same estimation sample and test sample. The results show SVM is much superior to WNN for small sample learning.

A new forecasting method of wind speed based on wavelet analysis and support vector machine is proposed in this paper. By the method, firstly, the original wind speed sequences are decomposed into coarse components and detail components, and secondly, every wavelet components are separately forecasted with corresponding support vector machine models. Finally, the forecasting results of original wind speed series are achieved by using wavelet reconstruction. The experimental results by simulation prove that this method is capable of improving generalization performance and forecasting precision. The mean absolute percentage error (MAPE) reaches 8.21%.

The automatic pattern recognition of power network fault is important work for power network fault diagnosis in power system. This paper presents automatic pattern recognition methods based on wavelet packet time entropy and support vector machine (SVM) for distribution network operation states. The proposed methods employ wavelet packet time entropy to extract the most important features in the current when the faults happen in the power network. Then the type of power network fault is identified by SVM. The proposed method is found the most suitable pattern recognition tool in the power network fault diagnosis by simulations in typical distribution network. The simulation results show that the method has good performance on pattern recognition with low dimension of eigenvector, less training samples and high classification accuracy.

With the imminent completion of the Human Genome Project and the fast increase of many complete genomes of prokaryotes and eukaryotes, the task of organizing and understanding the generated sequence and structural data becomes more and more pressing and demands better and efficient analysis algorithms. Computational gene identification is an issue of obvious importance as a tool of identifying biologically relevant features that often cannot be found by traditional sequence database searching technique. It bases on the accurate gene structure and developing accurate splice sites identification algorithms plays an important role in the identification of gene structure in eukaryote organisms. In this paper, we proposed a novel approach and developed an accurate splice sites identification algorithm based on discrete wavelet transform learned by support vector machine, which can achieve a superior performance for the prediction of splice sites, without any explicit use of local features. What's more we bring forward a new coding method for DNA sequence, so that the sequence data can be dealt with as image processing. The results show that our splice sites identification method is much better than or similar to other methods that are developed based on explicit statistical features, despite that it is so easy and simple. So this method could be applied in other signal detection problems for a try.

We proposed a new method of aero-engine early fault intelligent diagnosis which combined with stochastic resonance, wavelet packet analysis and support vector machine. This method can effectively extract the early fault feature of aero-engine and it can fast identify the early faults. At first, we use the principle of stochastic resonance to zooms the early weak fault feature signals and amplify fault features. Then, we make use of multi-resolution analysis characteristic of wavelet packet to extract the early fault feature vectors. At last, the feather vector is inputted to a classifier which is constructed by support vector machines and carries on identification of the early faults. The results shown that its effect of classification identification is well and it is effective to identify early fault in strong noise.

In this paper, a new image denoising method based on wavelet analysis and support vector machine regression (SVR) is presented. The feasibility of image denoising via support vector regression is discussed and is demonstrated by an illustrative example which denoise a 1-dimension signal with Gauss KBF SVM. The wavelet theory is discussed and applied to construct the wavelet kernel, then the wavelet support vector machine (WSVM) is proposed. The result of experiment shows that the denoising method based on WSVM can reduce noise well, the comparison between the method proposed in this paper and other ones is also given which proves this method is better than Gaussian KBF SVM and other traditional methods.

An original dual-class voting mechanism was put forward as the final decision method for brain computer interface, which is based on wavelet packet decomposition (WPD) to extract the brainwave features from EEG signals, and support vector machines (SVMs) to classify five mental tasks. Moreover, several preprocessing methods were applied efficiently. Segmentation along the time axis for increasing the correct classification rate, and nonlinear as well as linear normalization for emphasizing the important information in small magnitude and optimizing data distribution. Further, an especial grouping method was proposed to realize optimizing parameters automatically. Approximately, 95% of correct classification rate is obtained based on the proposed method, which is higher than the conventional.

The increased integration of wind power into the electric grid poses new challenges due to its fluctuation and volatility. Short term wind power forecasting is one of the most effective ways to deal with it. Various individual non-linear models are proposed to meet the data requirement to forecast short term wind power. However, as every model has its advantage and weakness, when these models are applied to different wind farms, the forecasting accuracy of every model varies because of distinct data character. This paper analyzes individual forecast models like Wavelet Transform and Support Vector Machine (SVM), and then puts forward a complex-valued forecasting model which is based on Artificial Natural Network in accordance with forecasting data provided by National Climatic Data Center in U.S. The existing individual models are matched and trained according to certain means by Natural Network to propose a multistage model. For variable data from different wind farms, the model can adjust and optimize portion of individual models. Compared with each single model, the multistage model has more robust adaptation and faster calculation speed, which can improve the forecasting precision and have more engineering value.

In this paper, iris recognition system using wavelet packet and support vector machines is presented. It specifically uses the multiresolution decomposition of 2-D discrete wavelet packet transform for extracting the unique features from the acquired iris image. This method of feature extraction is well suited to describe the shape of the iris while allowing the algorithm to be translation and rotation invariant. The SVM approach for comparing the similarity between the similar and different irises can be assessed to have the feature’s discriminative power. We have showed that the proposed method for human iris recognition gave a way of representing iris patterns in an efficient manner and thus had advantages of saving both time and space. Thanks to the efficiency of the proposed method, it can be easily applied to the real problems.

Partial discharge is an important cause of the deterioration of the cable insulation condition, and also an important manifestation of the deteriorated insulation condition. Partial discharge detection for cable insulation is an important means to find the cable insulation defects and assess state of the cable insulation. Partial discharge signals from different sources have great harm to the power equipment, so pattern identification is an important purpose in this research. First, the four kinds of discharge waveforms are processed by de-noising preprocessing, then discharge waveforms are decomposed by wavelet packet, and finally the energy percentage in different frequency bands can be obtained. Studies show that when the type of discharge signal is different, the wavelet packet coefficients in the distribution of each band is also different. So energy percentage of the wavelet packet coefficients can be used for discharge signal feature recognition. In this paper, SVM binary classification is first extended to multi-classification by M-ary algorithm, then the feature vectors composed by energy percentage of the wavelet packet coefficients are inputted into SVM for recognition. The recognition results of test samples show that feature vectors composed by the percentage of energy of wavelet packet coefficients can well reflect the characteristics of the original signal, and the method based on SVM has better recognition performance.

Short-Term Load Forecasting Based on Gaussian Wavelet SVM

A variety of academic theories are used to research the prediction algorithm including support vector machine. In order to improve instantaneity and accuracy of prediction, wavelet support vector machine prediction algorithm is studied, the structure and process of the algorithm is described also. Mackey-Glass <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">17</sub> time series is used as samples for the prediction result simulation, that is compared with the prediction effect used by basic support vector machine. The result shows that Root Mean Square Error of wavelet support vector machine prediction algorithm is 0.0075, that is superior to the basic support vector machine prediction algorithm.

Faced with the complexity of traditional multisensor information fusion and the low recognition rate of lower limb motion pattern, a lower limb motion pattern recognition method based on ReliefF(feature weighting algorithms) and KPCA(kernel principle component analysis) combined with SVM(support vector machine) is proposed. Firstly, IMU and plantar pressure were selected as the information sources for gait pattern recognition, and the time domain and energy features (mean value, mean square error, extreme value, correlation coefficient and wavelet entropy) of the collected lower limb motion information were selected for ReliefF feature selection. Secondly, KPCA feature dimensionality reduction is performed on the selected features to remove the redundant information, and the eigenvalues can reflect the characteristics of motion mode are obtained. Finally, SVM was used to classify the feature vectors of lower limb motion patterns, and 10 movement patterns were identified, such as walking, upstairs, downstairs, uphill, downhill, jogging, cycling, squatting, sitting and standing. The recognition results of this paper are compared with BP neural network and PNN probabilistic neural network, which verifies the feasibility and practicability of this method.

Epilepsy is the most common neurological disorder characterized by recurrent seizures due to the abnormality of brain neuronal discharge. One of the tools to help neurologists diagnose epilepsy relies on EEG signals extensively because it can represent the brain's neuron activity. Due to the EEG signal being considered a complex biological signal, the signal complexity-based feature extraction method is suitable for characterizing epileptic EEG signals. One method for extracting EEG signals is entropy which is able to quantify signal complexity. Several previously proposed methods were wavelet entropy and IMF entropy. This study proposed multi-distance signal level difference (MSLD) entropy as a feature extraction method to extract the EEG signal complexity. The decomposed signal from MSLD was then extracted using entropy measurement. In the classification process, a Support Vector Machine (SVM) was applied to classify the normal, pre-ictal, and ictal signals of EEG signals. The highest accuracy of 96% was obtained using cubic SVM as the classifier. The proposed method provides an alternative method for analyzing EEG or other biomedical signals based on signal complexity.

Pancreatic ductal adenocarcinoma (PDAC) is one of the most aggressive and lethal tumors worldwide, with limited effective treatments. Globally, the incidence of pancreatic cancer is expected to rise to 18.6 per 100,000 by 2050, with an average annual growth rate of 1.1%, implying that PDAC would represent a considerable public health burden. Identifying prognostic markers is critical for making therapy decisions and improving patient outcomes. In this study, the microarray gene expression data of PDAC were analyzed using artificial intelligence (AI) algorithms and molecular docking to identify the differentially expressed genes (DEGs) and drug repurposing. The GSE183795 dataset used in this study was obtained from the National Centre for Biotechnology Information. Further, the data were analyzed using GEO2R tools, and genes were selected based on logFC values>1. Then, these genes were ranked using AI algorithms such as support vector machine (SVM), logistic regression, random forest, extreme gradient boosting (XGB), and one-dimensional convolutional neural network to identify the DEGs. The performance of the models was evaluated using stratified 10-fold cross-validation and different classification metrics. A drug library was prepared using DepMap corresponding to the identified DEGs, and subsequently, molecular docking and pharmacokinetics analysis were performed. The result of the logFC>1 listed 107 upregulated genes in PDAC. It was observed that SVM and XGB show the average 10-fold accuracy, sensitivity, specificity, precision, and F-score of 79.25%, 78.37%, 78.37%, 79.33% and 78.35% respectively. Our results revealed that LIFR, BTG2, EPHX2, and PAK3 are within the top three and commonly ranked by AI models. Further, we identified three drugs, such as BI-2536, Ponatinib (AP-24534), and AZ-628, which show the best efficacy based on the binding energies by molecular docking analysis. The pharmacokinetics study strengthened our results that the identified drugs can be used as a therapeutic for PDAC as they obey Lipinski's rule. In conclusion, identified genes can act as prognostic markers, and drugs could be used as potential therapeutics for PDAC.