Optimal Short-Time Features for Music/Speech Classification of Compressed Audio Data

Feature extraction for machine learning-based intrusion detection in IoT networks

Remote sensing technology provides an unprecedented tool for the continuous and real-time monitoring of atmospheric SO2 from volcanic eruption and anthropogenic emission. The Global Ozone Monitoring Experiment (GOME), SCanning Imaging Absorption spectroMeter for Atmospheric CHartographY (SCIAMACHY), and Ozone Monitoring Instrument (OMI) have high SO2 monitoring capability. The OMI, which was launched on the EOS/Aura platform in July 2004, has the same hyperspectral measurements as the GOME and SCIAMACHY, but offers the improved spatial resolution at nadir (1324 km2) and daily global coverage for short-lifetime SO2. For OMI operational SO2 planetary boundary layer (PBL) retrieval, the previous band residual difference (BRD) algorithm has been replaced by principal component analysis (PCA) algorithm, which effectively reduces the systematic biases in SO2 column retrievals. However, there are few studies on the evaluations and validations of PCA SO2 retrievals over China, and the long-term comparisons with BRD SO2 retrievals also need to be conducted. In this study, the accuracies of PCA and BRD SO2 retrievals are validated by using ground-based multi axis differential optical absorption spectroscopy (MAX-DOAS) located in Beijing, and regional atmospheric modeling system, community multi-scale air quality (RAMS-CMAQ) modeling system model which can simulate the vertical distribution of atmospheric SO2. Moreover, BRD and PCA SO2 retrievals from oceanic area, eastern China and Reunion volcanic eruption are compared to find the long-term trend and spatiotemporal differences between SO2 columns. Finally, the uncertainty of SO2 retrieval, caused by measurement errors, band selection and input parameter errors in radiative transfer model, are analysed to understand the limitations of BRD and PCA algorithms. Results show that both PCA and BRD SO2 retrievals over Beijing are lower than ground-based MAX-DOAS measurements of SO2. PCA and BRD SO2 retrievals over eastern China are lower than the simulated SO2 columns from RAMS-CMAQ in winter 2008, but in July and August BRD SO2 columns are higher than RAMS-CMAQ simulations. The values of SO2 columns from BRD over China are more consistent with those from ground-based MAX-DOAS and RAMS-CMAQ model than from PCA. Although PCA algorithm effectively reduces the noise in SO2 column retrieval, SO2 columns from PCA over China are lower than those from BRD. For oceanic area where SO2 amount is nearly zero, the standard deviation of PCA results is lower than that of BRD, but the absolute value of averaged PCA SO2 column is larger than that of BRD. In the case of Reunion volcanic eruption with SO2 columns larger than 25 DU, the BRD SO2 columns are lower than PCA retrievals. Meanwhile, with the increase of SO2 column, the difference between BRD and PCA SO2 retrievals increases. Detailed uncertainty analysis shows the influences of measurement errors, band selection and inputs of radiative transfer model on the retrieval results. This study is important for developing the retrieval algorithm, and can also improve the application of OMI SO2 products.

Recognizing human emotions by machines is a complex task. Deep learning models attempt to automate this process by rendering machines to exhibit learning capabilities. However, identifying human emotions from speech with good performance is still challenging. With the advent of deep learning algorithms, this problem has been addressed recently. However, most research work in the past focused on feature extraction as only one method for training. In this research, we have explored two different methods of extracting features to address effective speech emotion recognition. Initially, two-way feature extraction is proposed by utilizing super convergence to extract two sets of potential features from the speech data. For the first set of features, principal component analysis (PCA) is applied to obtain the first feature set. Thereafter, a deep neural network (DNN) with dense and dropout layers is implemented. In the second approach, mel-spectrogram images are extracted from audio files, and the 2D images are given as input to the pre-trained VGG-16 model. Extensive experiments and an in-depth comparative analysis over both the feature extraction methods with multiple algorithms and over two datasets are performed in this work. The RAVDESS dataset provided significantly better accuracy than using numeric features on a DNN.

In the last decade there has been a considerable development of spaceborne SAR sensors. All the major space agencies are planning future SAR missions with polarimetric capabilities. However there is still a need to guide electromagnetic and statistics theories that take advantage of this kind of information towards operational applications. The use of contextual information is often required for automatic interpretation and target detection. The implementation of fast and reliable algorithms that exploit both polarimetric and contextual information can be limited by the increased dimensionality of the problem. Principal Component Analysis (PCA) is a data analysis technique that relies on a simple transformation of recorded observation, stored in a vector, to produce statistically independent variables. Non-Linear PCA is commonly seen as a non-linear generalization and extension of standard PCA. If non-linear correlations between variables exist, NLPCA will describe the data with greater accuracy and/or by fewer factors than PCA. In this work a combination of polarimetric and contextual information is performed using an Auto Associative Neural Network. A set of polarimetric input features were chosen together with contextual descriptors in order to produce an information set having lower dimensionality that can be exploited in a classification problem.

In this study, an iterative k-t principal component analysis (PCA) algorithm with nonrigid frame-to-frame motion correction is proposed for dynamic contrast-enhanced three-dimensional perfusion imaging. An iterative k-t PCA algorithm was implemented with regularization using training data corrected for frame-to-frame motion in the x-pc domain. Motion information was extracted using shape-constrained nonrigid image registration of the composite of training and k-t undersampled data. The approach was tested for 10-fold k-t undersampling using computer simulations and in vivo data sets corrupted by respiratory motion artifacts owing to free-breathing or interrupted breath-holds. Results were compared to breath-held reference data. Motion-corrected k-t PCA image reconstruction resolved residual aliasing. Signal intensity curves extracted from the myocardium were close to those obtained from the breath-held reference. Upslopes were found to be more homogeneous in space when using the k-t PCA approach with motion correction. Iterative k-t PCA with nonrigid motion correction permits correction of respiratory motion artifacts in three-dimensional first-pass myocardial perfusion imaging.

In this paper, Arif Index is proposed that can be used to assess the discrimination power of features in pattern classification problems. Discrimination power of features play an important role in the classification accuracy of a particular classifier applied to the pattern classification problem. Optimizing the performance of a classifier requires a prior knowledge of maximum achievable accuracy in pattern classification using a particular set of features. Moreover, it is also desirable to know that this set of features is separable by a decision boundary of any arbitrary complexity or not. Proposed index varies linearly with the overlap of features of different classes in the feature space and hence can be used in predicting the classification accuracy of the features that can be achieved by some optimal classifier. Using synthetic data, it is shown that the predicted accuracy and Arif index are very strongly correlated with each other (R 2 = 0.99). Implementation of the index is simple and time efficient. Index was tested on Arrhythmia beat classification problem and predicted accuracy was found to be in consistent with the reported results.

In this paper, we propose a new spectral-spatial classification strategy to enhance the classification performances obtained on hyperspectral images by integrating rotation forests and Markov random fields (MRFs). First, rotation forests are performed to obtain the class probabilities based on spectral information. Rotation forests create diverse base learners using feature extraction and subset features. The feature set is randomly divided into several disjoint subsets; then, feature extraction is performed separately on each subset, and a new set of linear extracted features is obtained. The base learner is trained with this set. An ensemble of classifiers is constructed by repeating these steps several times. The weak classifier of hyperspectral data, classification and regression tree (CART), is selected as the base classifier because it is unstable, fast, and sensitive to rotations of the axes. In this case, small changes in the training data of CART lead to a large change in the results, generating high diversity within the ensemble. Four feature extraction methods, including principal component analysis (PCA), neighborhood preserving embedding (NPE), linear local tangent space alignment (LLTSA), and linearity preserving projection (LPP), are used in rotation forests. Second, spatial contextual information, which is modeled by MRF prior, is used to refine the classification results obtained from the rotation forests by solving a maximum a posteriori problem using the α-expansion graph cuts optimization method. Experimental results, conducted on three hyperspectral data with different resolutions and different contexts, reveal that rotation forest ensembles are competitive with other strong supervised classification methods, such as support vector machines. Rotation forests with local feature extraction methods, including NPE, LLTSA, and LPP, can lead to higher classification accuracies than that achieved by PCA. With the help of MRF, the proposed algorithms can improve the classification accuracies significantly, confirming the importance of spatial contextual information in hyperspectral spectral-spatial classification.

Sensor nodes fault detection for agricultural wireless sensor networks based on NMF

This paper characterizes an approach to data compression and defect isolation for the multi-variable process by introducing the wavelet principal component analysis (PCA) and the genetic algorithms. In the defect analysis process, data compression is a critical phase. In our case of study, this can be achieved through the wavelet PCA. We can, through the use of such a technique, filter the data noise and the measurement errors by combining the PCA and the wavelet analysis. The genetic algorithm optimization feature is then employed in structuring the residues of the defect isolation. We present the principle of the PCA as well as the wavelet compression property. Then, we describe both the defect isolation classical method by structuring the residues and the principle of optimizing a problem by the genetic algorithms. The proposed approach is implemented on a statistic system and the Tennessee Eastman process. In comparison with other previous methods, the obtained results mirror the performance of the approach to defect isolation.

Face recognition is one technology that is commonly used now. Therefore, various algorithms continue to be developed to obtain maximum results with minimum costs. One of them is the Deep Neural Network or DNN algorithm. While DNN requires a large dataset to train the algorithm, another algorithm called the Principal Component Analysis (PCA) algorithm works good in a smaller dataset. These algorithms are compared to know which algorithm has the better result in given circumstances. Later the accuracy, speed, and optimality of the algorithms are analyzed. By comparing these algorithms, we could know which algorithm is preferable in given circumstances. After the experiment is done, we can assume that the two algorithms have a slight difference in terms of accuracy. Also, the time used for running the PCA implementation code is slightly longer than DNN. However, that does not mean that the PCA algorithm is not great. If the dataset to be used is limited, PCA is going to be a good choice.

Cancer is the most common disease to threaten human health. The ability to screen individuals with malignant tumours with only a blood sample would be greatly advantageous to early diagnosis and intervention. This study explores the possibility of discriminating between cancer patients and normal subjects with serum surface-enhanced Raman spectroscopy (SERS) and a support vector machine (SVM) through a peripheral blood sample. A total of 130 blood samples were obtained from patients with liver cancer, colonic cancer, esophageal cancer, nasopharyngeal cancer, gastric cancer, as well as 113 blood samples from normal volunteers. Several diagnostic models were built with the serum SERS spectra using SVM and principal component analysis (PCA) techniques. The results show that a diagnostic accuracy of 85.5% is acquired with a PCA algorithm, while a diagnostic accuracy of 95.8% is obtained using radial basis function (RBF), PCA–SVM methods. The results prove that a RBF kernel PCA–SVM technique is superior to PCA and conventional SVM (C-SVM) algorithms in classification serum SERS spectra. The study demonstrates that serum SERS, in combination with SVM techniques, has great potential for screening cancerous patients with any solid malignant tumour through a peripheral blood sample.

Aiming at the accuracy of continuous monitoring of blood pressure by photoelectric method, and being unable to establish a unified prediction model for all individuals in the population due to the differences of human characteristics, this paper proposes a blood pressure prediction method based on principal component analysis (PCA) and genetic algorithm (GA) to optimize machine learning model. The method firstly processes the photoplethysmography (PPG) signal, the electrocardiography (ECG) signal and the human body features to form a feature matrix, and uses a machine learning model to perform regression training on the feature matrix and the real-time blood pressure value measured by the mercury sphygmomanometer. Then, the GA is used to optimize the parameters to establish an optimal blood pressure prediction model. The experimental results show that compared with the traditional SVR, the proposed method could improve the predictive accuracy by 10%–15%.

Within the last years various principal component analysis (PCA) algorithms have been proposed. In this paper we use a general framework to describe those PCA algorithms which are based on Hebbian learning. For an important subset of these algorithms, the local algorithms, we fully describe their equilibria, where all lateral connections are set to zero and their local stability. We show how the parameters in the PCA algorithms have to be chosen in order to get an algorithm which converges to a stable equilibrium which provides principal component extraction.

The aim of this paper is to propose a real-time classification algorithm for the low-amplitude electroencephalography (EEG) signals, such as those produced by remembering an unpleasant odor, to drive a brain-computer interface. The peculiarity of these EEG signals is that they require ad hoc signals preprocessing by wavelet decomposition, and the definition of a set of features able to characterize the signals and to discriminate among different conditions. The proposed method is completely parameterized, aiming at a multiclass classification and it might be considered in the framework of machine learning. It is a two stages algorithm. The first stage is offline and it is devoted to the determination of a suitable set of features and to the training of a classifier. The second stage, the real-time one, is to test the proposed method on new data. In order to avoid redundancy in the set of features, the principal components analysis is adapted to the specific EEG signal characteristics and it is applied; the classification is performed through the support vector machine. Experimental tests on ten subjects, demonstrating the good performance of the algorithm in terms of both accuracy and efficiency, are also reported and discussed.

This article presents a method based on support vector machines (SVMs) and the Osuna–Platt algorithm used to diagnose the ischemic heart disease. It also includes the necessary concepts from the optimization theory which make it possible to formulate the problem for support vectors and the Osuna–Platt algorithm applied. Next, a principal component analysis (PCA) algorithm used is also presented. Heart images acquired using Single Photon Emission Computed Tomography (SPECT) have been used in the experimental part. Results of classifying cardiac SPECT images using SVM, PCA and neural networks are compared here with those obtained using another method of machine learning – CLIP3 – a combination of the decision tree algorithm and the rule induction algorithm. Tests against an SPECT image database have shown that SVMs are generally more accurate and specific, while the PCA algorithm is the most sensitive for all data sets analysed in this research project.