Extreme Leaning Machine Research Articles

Multilayer Fisher extreme learning machine for classification

As a special deep learning algorithm, the multilayer extreme learning machine (ML-ELM) has been extensively studied to solve practical problems in recent years. The ML-ELM is constructed from the extreme learning machine autoencoder (ELM-AE), and its generalization performance is affected by the representation learning of the ELM-AE. However, given label information, the unsupervised learning of the ELM-AE is difficult to build the discriminative feature space for classification tasks. To address this problem, a novel Fisher extreme learning machine autoencoder (FELM-AE) is proposed and is used as the component for the multilayer Fisher extreme leaning machine (ML-FELM). The FELM-AE introduces the Fisher criterion into the ELM-AE by adding the Fisher regularization term to the objective function, aiming to maximize the between-class distance and minimize the within-class distance of abstract feature. Different from the ELM-AE, the FELM-AE requires class labels to calculate the Fisher regularization loss, so that the learned abstract feature contains sufficient category information to complete classification tasks. The ML-FELM stacks the FELM-AE to extract feature and adopts the extreme leaning machine (ELM) to classify samples. Experiments on benchmark datasets show that the abstract feature extracted by the FELM-AE is more discriminative than the ELM-AE, and the classification results of the ML-FELM are more competitive and robust in comparison with the ELM, one-dimensional convolutional neural network (1D-CNN), ML-ELM, denoising multilayer extreme learning machine (D-ML-ELM), multilayer generalized extreme learning machine (ML-GELM), and hierarchical extreme learning machine with L21‑norm loss and regularization (H-LR21-ELM).

Multiple power quality disturbances analysis in photovoltaic integrated direct current microgrid using adaptive morphological filter with deep learning algorithm

In this paper, an adaptive multiscale enhanced average combination morphological filter is proposed to analyze voltage signals captured at the point of common coupling of the photovoltaic integrated direct current microgrid for diagnosing several power quality disturbances. This method performs multiscale operation on the power quality disturbance signals from which the optimized structuring element scales are chosen by using maximum value of sparse envelope kurtosis index. Further, statistical features are extracted by using optimized scales. The extracted features are used as inputs to a deep stacked extreme leaning machine autoencoder to retrieve the most distinguishable unsupervised features and these features are processed through kernel random vector functional link network based supervised classifier to classify the power quality disturbances. Further, to improve the classification performance of the proposed technique, a meta heuristic chaotic particle swarm optimization integrated with sine cosine optimization algorithm is used. This optimization algorithm optimizes the kernel parameters by minimizing the mean absolute error. The efficacy of the proposed technique is tested in terms of classification accuracy and confusion matrix and is verified by multifunction high-speed national instrument device for monitoring of power quality disturbances in the MATLAB/Simulink platform. The simulations and hardware results demonstrate that the proposed technique exhibits remarkable accuracy, faster training time and low computational complexity as compared with other existing methods.

Application of improved ELM algorithm in the prediction of earthquake casualties.

Earthquake casualties prediction is a basic work of the emergency response. Traditional forecasting methods have strict requirements on sample data and lots of parameters are required to be set manually, which can result in poor results with low prediction accuracy and slow learning speed. In this paper, the Extreme Leaning Machine (ELM) is introduced into the earthquake disaster casualty predictions with the purpose of improving the prediction accuracy. However, traditional ELM model still has the problems of poor network structure stability and low prediction accuracy. So an Adaptive Chaos Particle Swarm Optimization (ACPSO) is proposed to the optimize traditional ELM's network parameters to enhance network stability and prediction accuracy, and the improved ELM model is applied to earthquake disaster casualty prediction. The experimental results show that the earthquake disaster casualty prediction model based on ACPSO-ELM algorithm has better stability and prediction accuracy. ACPSO-ELM algorithm has better practicality and generalization in earthquake disaster casualty prediction.

Dissolved oxygen prediction using a new ensemble method.

Prediction of dissolved oxygen which is an important water quality (WQ) parameter is crucial for aquatic managers who have responsibility for the ecosystem health's maintenance and for the management of reservoirs related to WQ. This study proposes a new ensemble method, Bayesian model averaging (BMA), for estimating hourly dissolved oxygen. The potential of the BMA was investigated and compared with five data-driven methods, extreme leaning machine (ELM), artificial neural networks (ANNs), adaptive neuro-fuzzy inference system (ANFIS), classification and regression tree (CART), and multilinear regression (MLR), by considering hourly temperature, pH, and specific conductivity data as inputs. The methods were compared with respect to three statistics, root mean square errors (RMSE), Nash-Sutcliffe efficiency, and determination coefficient. Results based on two stations' data indicated that the proposed method performed superior to the ELM, ANN, ANFIS, CART, and MLR in estimation of hourly dissolved oxygen; corresponding improvements obtained by BMA are about 5-8%, 13-12%, 7-9%, and 18-27% with respect to RMSE. The ELM also outperformed the other four methods (ANN, ANFIS, CART, and MLR), and the CART and MLR indicated the lowest estimation accuracy in both stations. Examination of various input combinations revealed that the most effective variable is water temperature while the specific conductivity has negligible effect on hourly dissolved oxygen.

A Method to Measure Thermal Conductivity of Vacuum Insulation Panel Using Enhanced Extreme Learning Machine Model

Thermal conductivity is an important quantity which represents the characteristic of Vacuum Insulation Panel’s (VIP’s) performance. Precise measurement of thermal conductivity provides better quality assurance for the users. In this paper, we presented a novel embedded sensor method to measure the thermal conductivity of VIP. The proposed method evaluated the quality of VIP primarily based on the relationship between thermal conductivity and frequency characteristic of the output signal. In addition, we presented a new mean ridge regression extreme leaning machine (M-RRELM) model via improving extreme learning machine (ELM) by ridge regression to modify the relationship between the thermal conductivity and the output signal frequency characteristic. Experiments have shown that the M-RRELM model has higher precision compared with the traditional ELM. The proposed method achieved good performance and was faster than the well known methods.

A lightweight and cost effective edge intelligence architecture based on containerization technology

The integration of Cloud computing and Internet of Things led to rapid growth in the edge computing field. This would not be achievable without combining the data centers’ managing systems with much more restrained technologies. One significantly effective and lightweight solution to this issue is presented by the Docker technology. It is able to manage virtualization process and can therefore be used to distribute, deploy and manage cloud and edge applications assigned into the clusters. In our study, this technology was represented by the Raspberry Pi devices, which are convenient thanks to their low cost, robust applicability and lightweight nature. Our application scenario focuses on identification of the human activities. In this paper, we suggest and evaluate an architecture on the basis of the distributed edge/cloud integration paradigm. We explain all of its advantages which lie in the combination of affordability and several other benefits provided by the fact that data processing is conducted by the edge devices instead of the central server. To recognize and identify human activity, the Regularized Extreme Leaning Machine (RELM) was engaged in our architecture. Our study presents detailed information about our use case scenario and the experimental simulation we performed.

Identification of Alzheimer's disease and mild cognitive impairment using multimodal sparse hierarchical extreme learning machine.

Different modalities such as structural MRI, FDG-PET, and CSF have complementary information, which is likely to be very useful for diagnosis of AD and MCI. Therefore, it is possible to develop a more effective and accurate AD/MCI automatic diagnosis method by integrating complementary information of different modalities. In this paper, we propose multi-modal sparse hierarchical extreme leaning machine (MSH-ELM). We used volume and mean intensity extracted from 93 regions of interest (ROIs) as features of MRI and FDG-PET, respectively, and used p-tau, t-tau, and as CSF features. In detail, high-level representation was individually extracted from each of MRI, FDG-PET, and CSF using a stacked sparse extreme learning machine auto-encoder (sELM-AE). Then, another stacked sELM-AE was devised to acquire a joint hierarchical feature representation by fusing the high-level representations obtained from each modality. Finally, we classified joint hierarchical feature representation using a kernel-based extreme learning machine (KELM). The results of MSH-ELM were compared with those of conventional ELM, single kernel support vector machine (SK-SVM), multiple kernel support vector machine (MK-SVM) and stacked auto-encoder (SAE). Performance was evaluated through 10-fold cross-validation. In the classification of AD vs. HC and MCI vs. HC problem, the proposed MSH-ELM method showed mean balanced accuracies of 96.10% and 86.46%, respectively, which is much better than those of competing methods. In summary, the proposed algorithm exhibits consistently better performance than SK-SVM, ELM, MK-SVM and SAE in the two binary classification problems (AD vs. HC and MCI vs. HC).

Flow profile identification with multipath transducers

Accuracy of flow rate determination is very critical to an ultrasonic gas flowmeter, which is sensitive to the flow profile under measurement and cannot be obviously improved with traditional weighted integration methods for multipath transducers. Therefore, on one hand, more attentions are paid to intelligent learning algorithms (e.g. artificial neural networks) in recent studies to accurately construct the mapping relation between multipath velocities and the flow rate. However, the bottleneck that a trained network is only customized for a certain flow profile greatly restricts its application. On the other hand, many researchers turn to reconstruct the flow field but so far all flow visualization methods cost too much and are difficult to realize online. In this paper, an intelligent method of flow profile identification for multipath ultrasonic flowmeters is developed to solve the above predicament. Based on support vector machine, a multiclass identifier is constructed to automatically identify the flow profile to be measured from 65 typical flow patterns. Extremely high identification accuracy of 99.49% for test and 100% for training is achieved while the test result still reaches 88.46% even when a measurement uncertainty of ±2% is considered. A comparison with extreme leaning machine further reveals the identification performance and robustness of the proposed flow profile identifier. Different piping configurations, installation positions and angles can be therefore accurately identified, which indirectly realizes the flow visualization and can comprise an intelligent system of ultrasonic flow measurement when combined with the intelligent flow rate determination algorithms.

State Preserving Extreme Learning Machine: A Monotonically Increasing Learning Approach

Extreme Learning Machines (ELM) has been introduced as a new algorithm for training single hidden layer feedforward neural networks instead of the classical gradient-based approaches. Based on the consistency property of data, which enforces similar samples to share similar properties, ELM is a biologically inspired learning algorithm that learns much faster with good generalization and performs well in classification tasks. However, the stochastic characteristics of hidden layer outputs from the random generation of the weight matrix in current ELMs leads to the possibility of unstable outputs in the learning and testing phases. This is detrimental to the overall performance when many repeated trials are conducted. To cope with this issue, we present a new ELM approach, named State Preserving Extreme Leaning Machine (SPELM). SPELM ensures the overall training and testing performance of the classical ELM while monotonically increases its accuracy by preserving state variables. For evaluation, experiments are performed on different benchmark datasets including applications in face recognition, pedestrian detection, and network intrusion detection for cyber security. Several popular feature extraction techniques, namely Gabor, pyramid histogram of oriented gradients, and local binary pattern are also incorporated with SPELM. Experimental results show that our SPELM algorithm yields the best performance on tested data over ELM and RELM.

Thermal sensation prediction by soft computing methodology

Thermal comfort in open urban areas is very factor based on environmental point of view. Therefore it is need to fulfill demands for suitable thermal comfort during urban planning and design. Thermal comfort can be modeled based on climatic parameters and other factors. The factors are variables and they are changed throughout the year and days. Therefore there is need to establish an algorithm for thermal comfort prediction according to the input variables. The prediction results could be used for planning of time of usage of urban areas. Since it is very nonlinear task, in this investigation was applied soft computing methodology in order to predict the thermal comfort. The main goal was to apply extreme leaning machine (ELM) for forecasting of physiological equivalent temperature (PET) values. Temperature, pressure, wind speed and irradiance were used as inputs. The prediction results are compared with some benchmark models. Based on the results ELM can be used effectively in forecasting of PET.

An Online Learning Model of Mobile User Preference Based on Context Quantification

In mobile network, the mobile user has the strict requirement for the performance of accessing the information. In order to provide the personalized service for mobile user timely and accurately, an online learning model of mobile user preference based on context quantification is proposed. In the model, a context quantification method is proposed, which can enhance the accuracy of learned mobile user preference; and the sliding window and the online extreme leaning machine (O-ELM) are introduced to realize the online learning. Firstly, it needs to judge whether the mobile user preference is affected by the context through analyzing the mobile user behaviors. Secondly, the context is quantified according to the context relevancy and the context similarity. And then, the sliding window is employed to select the samples that need to be learned when updating the mobile user preference. Finally, O-ELM is employed to learn the mobile user preference. The experimental results show that the proposed method surpasses the existing methods in the performance.

Meta-ELM: ELM with ELM hidden nodes

Extreme Leaning Machine (ELM) simply randomly assigns input weights and biases, ineluctably leading to certain stochastic behaviors and reducing generalization performance. In this paper, we propose a meta-learning model of ELM, called Meta-ELM. The Meta-ELM architecture consists of several base ELMs and one top ELM. Therefore, the Meta-ELM learning proceeds in two stages. First, each base ELM is trained on a subset of the training data. Then, the top ELM is learned with the base ELMs as hidden nodes. Theoretical analysis and experimental results on a few artificial and benchmark regression datasets show that the proposed Meta-ELM model is feasible and effective.

The study of using an extreme learning machine for rapid concentration estimation in multi-component gas mixtures

Cross-sensitivity is one of the major unpleasant characteristics of metal oxide gas sensors. In order to solve this challenging problem, artificial neural networks have proved to be very powerful tools, among which back propagation (BP) and radial basis function (RBF) neural networks are the two most commonly used ones in data analysis of metal oxide gas sensors and arrays. However, relatively long training time is the major disadvantage for the BP and RBF neural networks. In order to solve this problem, an extreme leaning machine (ELM) is introduced and studied in this paper. Experimental results show that ELM networks can achieve 466 and 333 times faster training speed than the BP and RBF neural networks, respectively. In addition, ELM networks can achieve comparable concentration prediction accuracy to RBF networks which is much better than BP networks.