Extreme learning machine with multi-scale local receptive fields for texture classification

Extreme learning machine (ELM), which was originally proposed for "generalized" single-hidden layer feedforward neural networks (SLFNs), provides efficient unified learning solutions for the applications of feature learning, clustering, regression and classification. Different from the common understanding and tenet that hidden neurons of neural networks need to be iteratively adjusted during training stage, ELM theories show that hidden neurons are important but need not be iteratively tuned. In fact, all the parameters of hidden nodes can be independent of training samples and randomly generated according to any continuous probability distribution. And the obtained ELM networks satisfy universal approximation and classification capability. The fully connected ELM architecture has been extensively studied. However, ELM with local connections has not attracted much research attention yet. This paper studies the general architecture of locally connected ELM, showing that: 1) ELM theories are naturally valid for local connections, thus introducing local receptive fields to the input layer; 2) each hidden node in ELM can be a combination of several hidden nodes (a subnetwork), which is also consistent with ELM theories. ELM theories may shed a light on the research of different local receptive fields including true biological receptive fields of which the exact shapes and formula may be unknown to human beings. As a specific example of such general architectures, random convolutional nodes and a pooling structure are implemented in this paper. Experimental results on the NORB dataset, a benchmark for object recognition, show that compared with conventional deep learning solutions, the proposed local receptive fields based ELM (ELM-LRF) reduces the error rate from 6.5% to 2.7% and increases the learning speed up to 200 times.

Lately, 3D imaging techniques have achieved a lot of progress due to recent developments in 3D sensor technologies. This leads to a great interest regarding 3D image feature extraction and classification techniques. As pointed out in literature, one of the most important and discriminative features in images is the textural content. Within this context, we propose a texture feature extraction technique for volumetric images with improved discrimination power. The method could be used in textured volumetric data classification tasks. To achieve this, we fuse two complementary pieces of information, feature vectors derived from Local Binary Patterns (LBP) and the Gray-Level Co-occurrence Matrix-based methods. They provide information regarding the image pattern and the contrast, homogeneity and local anisotropy in the volumetric data, respectively. The performance of the proposed technique was evaluated on a public dataset consisting of volumetric textured images affected by several transformations. The classifiers used are the Support Vector Machine, k-Nearest Neighbours and Random Forest. Our method outperforms other handcrafted 3D or 2D texture feature extraction methods and typical deep-learning networks. The proposed technique improves the discrimination power and achieves promising results even if the number of images per class is relatively small.

Texture and its analysis methods are crucial for image feature extraction and classification. Digital elevation model (DEM) is the most important data source of digital terrain analysis and landform classification, and considerable research values are gained from texture feature extraction and analysis from DEM data. In this research, on the basis of optimal texture feature extraction, the hilly area in Shandong, China, was selected as the study area, and DEM data with a resolution of 500 m were used as the experimental data for landform classification. First, second-order texture measures and texture image were extracted from DEM data by using a gray level cooccurrence matrix (GLCM). Second, the variation characteristics of each texture measure were analyzed, and the optimal feature parameters, such as direction, gray level, and texture window, were determined. Meanwhile, the texture feature value, combined with maximum information, was calculated, and the multiband texture image was obtained by resolving three optimal texture measure images. Finally, a support vector machine (SVM) method was adopted to classify landforms on the basis of the multiband texture image. Results indicated that the texture features of DEM data can be sufficiently represented and measured via the quantitative GLCM method. However, the feature parameters during the texture feature value calculation required further optimization. Based on the image texture from DEM data, efficient classification accuracy and ideal classification effect were achieved.

Since the mine microseismic and blasting signals are non-linear and non-stationary, the traditional linear analysis methods cannot effectively classify the microseismic and blasting signals. In this study, a method for feature extraction and classification of mine microseismic signals based on CEEMDAN_SE (Complete Ensemble Empirical Mode Decomposition with Adaptive Noise_Sample Entropy) is proposed. Firstly, the microseismic of rock mass and blasting vibration signals are decomposed by CEEMDAN. Then five sensitive IMF components of IMF4~IMF8 (microseismic) and IMF1~IMF5 (blasting) are screened out by using correlation coefficient. Next, the sample entropy values of the sensitive IMF components are calculated. Finally, the sample entropy is used as the eigenvector and input into ELM (extreme learning machine) to classify and identify the microseismic and blasting signals. The results show that there is a significant difference between the sample entropy values of microseismic and blasting signals. Combining CEEMDAN_SE with ELM, the accuracy of classification, recognition of microseismic and blasting signals is 91.5%. Compared with EEMD_SE_ELM and EMD_SE_ELM, the classification, recognition accuracy of CEEMDAN_ SE_ELM is improved by 2.5% and 5%, respectively. Therefore, this method provides a new idea for the classification and recognition of mine microseismic signals.

Exploring the complementarity of THz pulse imaging and DCE-MRIs: Toward a unified multi-channel classification and a deep learning framework

Expected energy-based restricted Boltzmann machine for classification

In recent years use of image processing techniques for texture analysis of machined surface is gaining importance in the field of manufacturing. This manuscript addresses texture identification methodology using Wavelet transform and artificial intelligence techniques. Captured images of machined surface using Electric discharge machining, milling, sand blasting and shaping is decomposed in to sub images and then discrete wavelet transform is applied on the sub images. To select the base wavelet minimum permutation entropy criterion is applied and statistical features were calculated from the base wavelet. Training and testing of feature vector is performed using two artificial intelligence techniques support vector machine and artificial neural network for identifying textured surface images.100 % training identification of textured images is obtained using support vector machine and artificial neural network and 87.5 % and 100 % testing identification of textured images is obtained using support vector machine and artificial neural network respectively. Results revealed that the present methodology identifies machined surface images with high accuracy.

The study was intended to eliminate the noise in three-dimensional transvaginal ultrasound (3D-TVS) images and improve the diagnostic accuracy in intrauterine adhesion (IUA). The extreme learning machine (ELM) algorithm was introduced first for statement. One hundred and thirty cases of suspected IUA patients were taken as the research subjects. The denoising effects of ELM algorithm were evaluated in terms of mean square errors (MSE), peak signal-to-noise ratio (PSNR), and running time, and its diagnostic efficiency of IUA was identified from precise, specificity, and sensitivity. Furthermore, the support vector machine (SVM) algorithm was introduced for comparison. It was found that the MSE and PSNR of the ELM algorithm were 0.0021 and 64.5, respectively, and its average operation time was 11.22 ± 0.89s, that the MSE values of SVM algorithm and ELM algorithm were 0.0045 and 0.0021 and the PSNR values were 52.3 and 64.5, respectively, and that the average running time of SVM algorithm was 16.35 ± 1.33s, and the average running time of ELM algorithm was 11.22 ± 0.89s, superior to the SVM algorithm in denoising effects. Moreover, the ELM algorithm showed excellent diagnostic efficiency for patients with various degrees of IUA. In conclusion, ELM can effectively eliminate noise in 3D-TVS images and demonstrates excellent diagnostic efficiency on IUA, which is worthy of clinical application.

Epilepsy is one of the frequent brain disorder that may consequence in the brain dysfunction and cognitive disorders. Epileptic seizures can occur due to transient and unexpected electrical interruptions of brain. EEG (ElectroEncephaloGram) is one of the non-invasive methods for analyzing the human brain dynamics that affords a direct evaluation of cortical behavior. Seizures are featured by short and episodic neuronal synchronous discharges with considerably enlarged amplitude. This uneven synchrony may happen in the brain accordingly i.e., partial seizures visible only in few channels of the EEG signal or generalized seizures, which are seen in every channel of the EEG signal involving the whole brain. Existing analysis of epilepsy depends on difficult visual screening by extremely trained clinicians. Data recordings create very lengthy data, and therefore the inspection and identification of epilepsy takes more time for diagnosis. At present time computerized systems are usually established to make the diagnosis simpler. This paper discusses an implementation of automated epileptic EEG detection system using neural networks. In this paper, a statistical parameter regarded as Sample Entropy (SampEn), is used as a method for feature extraction for performing the task of classifying EEG signals, which are normal, ictal and interictal. Here the value of SampleEn plays a vital role in the diagnosis of Epileptic seizures. It is been observed that the value of SampEn falls suddenly during an Epileptic Seizure, and hence this information delineates the exact occurrence of Epilepsy. Classification strategies are based on models like Back-propagation Neural Network (BPNN),Support Vector Machine(SVM) and Extreme Learning Machine (ELM). Hybrid ELM (HELM) is the classification method devised in the current paper and is the first of its kind to be implemented for EEG classification. This classification method uses the Analytical Hierarchy Process (AHP) method to select the input weights and hidden biases, the ELM algorithm to analytically determine the output weights and the Levenberg-Marquardt (LM) algorithm to learn the network. Experimental results show that automatic epilepsy detection using sample entropy (SampEn) and Hybrid ELM achieves better accuracy in lesser time.

Malware forensics is a field in digital forensics that is dedicated to collecting evidence of a malware attack on a system. Malware is also executable code, but unlike regular executables, it is designed to perform unauthorized actions on victim computers. Being an executable, the malware also performs designated tasks through several processes that can leave its traces (digital footprints) throughout different components of the operating system, such as registries, file systems, DLL calls, API calls, memory usage, CPU utilization, network traffic, etc. Some of these traces can exist even after the removal of the malware and can be used as evidence to verify the occurrence of an attack. The proposed malware forensic analysis technique combines both the static and dynamic features of the malware to trace its evidence. This information is collected using the Cuckoo sandbox, which executes the malware sample files in a virtual environment to generate a JSON-formatted analysis report. The Python script is used to extract the non-volatile features from the JSON format analysis report. However, combining both features results in a very long and sparse feature vector, which negatively affects the classifier’s performance. Therefore, the principal component analysis (PCA) algorithm is adopted to reduce the dimensionality of the feature vectors. The generated reduced dimensionality feature vectors are used to train the extreme learning machine (ELM). The ELM can be trained in a fraction of the time compared to the support vector machine (SVM), while achieving similar accuracy. This makes it a favorable choice, especially for malware forensic analysis, where the classifier needs to train with a large amount of data and requires quick decision-making capabilities. The performance of the proposed techniques is evaluated using 90 malicious code samples containing 41 Trojans, 28 worms, and 21 bots. To demonstrate the superiority of the proposed algorithm, a comparison with some state-of-the-art classification methods, named Hidden Markov Model (HMM), Support Vector Machine (SVM), Artificial Neural Network (ANN), and ELM, was also performed. The comparison results show that the proposed technique achieves approximately 11% higher precision, 10% higher recall, and a 10.5% higher F-score than the competitors.

Various myoelectric prostheses controlled by electromyography (EMG) signals have been developed. However, there have been few studies that provide fast and accurate methods to predict handgrip force from EMG signals. Rapid and precise handgrip force prediction is required, especially for the real-time control system of myoelectric prostheses. In this study, extreme learning machine (ELM) is applied to predict handgrip force from surface EMG signals of forearm muscles. Furthermore, ELM is compared with support vector machine (SVM) and multiple nonlinear regression (MNLR). The below 10 % of the surface EMG and handgrip force signals were cut away, and then the root mean square feature extracted from the modified surface EMG signals was taken as input vector for these three kinds of predicting mechanisms. For the testing dataset, ELM achieved a slightly larger root mean squared error than SVM did and a smaller one than MNLR did. Meanwhile, all three methods showed high correlation coefficients. For the total processing time, ELM and MNLR consumed much less time than SVM did. Experimental results demonstrate that ELM possesses a relatively good accuracy and little consumed time, although SVM is effective for handgrip force estimation in terms of accuracy. Overall, ELM has a promising potential for predicting handgrip force rapidly and precisely.

The method of texture feature extraction and classification of pork loin B ultrasound image is proposed, which can be applied to the computer-aided judgment the pork loin fat content of pork loin. 5 texture features which is based on the texture of the co-occurrence matrix are extracted from the B ultrasound image of pork loin according to the digital image processing algorithm. Using the correlation analysis method to select the key texture extraction in the first step. Then,the classification is realized based on the BP neural network. The train set and test set are randomly chosen from 135 cases. Tests performed show that the proposed method result in a high classification accuracy, which will provide the researcher a valuable opinion on the pork fat content detection.

The purpose of this study is to develop a dissimilarity measure for the classification of trabecular bone (TB) texture in knee radiographs. Problems associated with the traditional extraction and selection of texture features and with the invariance to imaging conditions such as image size, anisotropy, noise, blur, exposure, magnification, and projection angle were addressed. In the method developed, called a signature dissimilarity measure (SDM), a sum of earth mover's distances calculated for roughness and orientation signatures is used to quantify dissimilarities between textures. Scale-space theory was used to ensure scale and rotation invariance. The effects of image size, anisotropy, noise, and blur on the SDM developed were studied using computer generated fractal texture images. The invariance of the measure to image exposure, magnification, and projection angle was studied using x-ray images of human tibia head. For the studies, Mann-Whitney tests with significance level of 0.01 were used. A comparison study between the performances of a SDM based classification system and other two systems in the classification of Brodatz textures and the detection of knee osteoarthritis (OA) were conducted. The other systems are based on weighted neighbor distance using compound hierarchy of algorithms representing morphology (WND-CHARM) and local binary patterns (LBP). Results obtained indicate that the SDM developed is invariant to image exposure (2.5-30 mA s), magnification (x1.00 - x1.35), noise associated with film graininess and quantum mottle (< 25%), blur generated by a sharp film screen, and image size (> 64 x 64 pixels). However, the measure is sensitive to changes in projection angle (> 5 degrees), image anisotropy (> 30 degrees), and blur generated by a regular film screen. For the classification of Brodatz textures, the SDM based system produced comparable results to the LBP system. For the detection of knee OA, the SDM based system achieved 78.8% classification accuracy and outperformed the WND-CHARM system (64.2%). The SDM is well suited for the classification of TB texture images in knee OA detection and may be useful for the texture classification of medical images in general.

Texture is an important significant property of medical images based on which images can be characterized and classified in a content-based image retrieval and classification system. This paper examines the feature extraction methods to ameliorate texture recognition accuracy by extracting the rotation-invariant texture feature from liver images by the individual Gabor filter method and by multi-scale Gabor rotation-invariant LBP (MGRLBP) method. The features extracted from both the approaches are tested on a set of 60 liver images of four different classes. The classification algorithms such as support vector machine (SVM) and k-nearest neighbor (KNN) were used to evaluate the extracted features from both methods, showing advancing improvements with the MGRLBP method over the individual method in the classification task.

A novel approach is described for the supervised classification of marble textures in different classes according to visual appearance, using sum and difference histograms for texture analysis and feature extraction, and support vector machines for classification. Results show very good discrimination between classes.