Abstract: Breast cancer plays a significant role in affecting female mortality. Researchers are actively seeking to develop early detection methods of breast cancer. Several technologies contributed to the reduction in mortality rate from this disease, but early detection contributes most to preventing disease spread, breast amputation and death. Thermography is a promising technology for early diagnosis where thermal cameras employed are of high resolution and sensitivity. The combination of Artificial Intelligence (AI) with thermal images is an effective tool to detect early stage breast cancer and is foreseen to provide impressive predictability levels. This paper reviews systematically the related works employing thermography with AI highlighting their contributions and drawbacks and proposing open issues for research. Several different types of Artificial Neural Networks (ANNs) and deep learning models were used in the literature to process thermographic images of breast cancer, such as Radial Basis Function Network (RBFN), K-Nearest Neighbors (KNN), Probability Neural Network (PNN), Support Vector Machine (SVM), ResNet50, SeResNet50, V Net, Bayes Net, Convolutional Neural Networks (CNN), Convolutional and DeConvolutional Neural Networks (C-DCNN), VGG-16, Hybrid (ResNet-50 and V-Net), ResNet101, DenseNet and InceptionV3. Previous studies were found limited to varying the numbers of thermal images used mostly from DMR-IR database. In addition, analysis of the literature indicate that several factors do affect the performance of the Neural Network used, such as Database, optimization method, Network model and extracted features. However, due to small sample size used, most of the studies achieved a classification accuracy of 80% to 100%.

Abstract: Radial Basis Function (RBF) networks are a classical family of algorithms for supervised learning. The most popular approach for training RBF networks has relied on kernel methods using regularization based on a norm in a Reproducing Kernel Hilbert Space (RKHS), which is a principled and empirically successful framework. In this paper we aim to revisit some of the older approaches to training the RBF networks from a more modern perspective. Specifically, we analyze two common regularization procedures, one based on the square norm of the coefficients in the network and another one using centers obtained by $k$ k -means clustering. We show that both of these RBF methods can be recast as certain data-dependent kernels. We provide a theoretical analysis of these methods as well as a number of experimental results, pointing out very competitive experimental performance as well as certain advantages over the standard kernel methods in terms of both flexibility (incorporating of unlabeled data) and computational complexity. Finally, our results shed light on some impressive recent successes of using soft $k$ k -means features for image recognition and other tasks.

Abstract: The continuous increase in energy consumption has brought worldwide attention to its significant environmental effect, which is triggered by the increase in greenhouse gas emissions, global warming, and rapid climate change. As such, more energy efficient buildings are required to minimize the energy consumption of heating and cooling. The present study introduces a set of machine learning-based models to predict the heating and cooling loads in buildings. This includes back-propagation artificial neural network, generalized regression neural network, radial basis neural network, radial kernel support vector machines and ANOVA kernel support vector machines. The comparisons were conducted as per mean absolute percentage error (MAPE), mean absolute error (MAE), relative absolute error (RAE), root relative squared error (RRSE) and root-mean squared error (RMSE). The significances of the capacities of the machine learning models are evaluated using two-tailed student’s t-tests. Eventually, a holistic evaluation of the machine learning models is conducted using average ranking algorithm. Results demonstrate that the radial basis function network outperformed the afore-mentioned machine learning models significantly.

Abstract: Improvement in area of artificial intelligence for predicting different hydrological phenomenon has shaped an enormous alteration in predictions. Knowledge on suspended sediment load (SSL) is vital in managing water resources problems and safe guard environment. Present study evaluated accurateness of five soft computing techniques, i.e. radial basis function network (RBFN), cascade forward back propagation neural network (CFBPNN), support vector machine (SVM), integration of support vector machine with firefly algorithm (SVM-FFA) and phase space reconstruction (PSR) with SVM-FFA (PSR-SVM-FFA) approaches to estimate daily SSL in Salebhata, Suktel, Lant gauge stations in western part of Odisha, India. Performance of selected models were evaluated on basis of performance criterion namely root mean square error (RMSE), Nash-Sutcliffe (NSE), Wilton index (WI) for choosing best fit model. Results acquired verified that application of various neural network methods in present field of study showed fine concurrence with observed SSL values. Comparison of estimation accuracies of different methods exemplified that PSR-SVM-FFA is very precise to estimate SSL when compared with other models. Result shows that Suktel gauge station, the best value of WI is 0.978 for PSR-SVM-FFA model, while it is 0.959, 0.923, 0.885, and 0.842 for SVM-FFA, SVM, CFBPNN, RBFN models in testing phase. Moreover, cumulative SSL data calculated by PSR-SVM-FFA method are closer to observed data as compared to other methods.

Abstract: The recent developments of computer and electronic systems have made the use of intelligent systems for the automation of agricultural industries. In this study, the temperature variation of the mushroom growing room was modeled by multi-layered perceptron and radial basis function networks based on independent parameters including ambient temperature, water temperature, fresh air and circulation air dampers, and water tap. According to the obtained results from the networks, the best network for MLP was in the second repetition with 12 neurons in the hidden layer and in 20 neurons in the hidden layer for radial basis function network. The obtained results from comparative parameters for two networks showed the highest correlation coefficient (0.966), the lowest root mean square error (RMSE) (0.787) and the lowest mean absolute error (MAE) (0.02746) for radial basis function. Therefore, the neural network with radial basis function was selected as a predictor of the behavior of the system for the temperature of mushroom growing halls controlling system.

Abstract: The paper discusses an adaptive strategy to effectively control nonlinear manipulation motions of a dual arm robot (DAR) under system uncertainties including parameter variations, actuator nonlinearities and external disturbances. It is proposed that the control scheme is first derived from the dynamic surface control (DSC) method, which allows the robot’s end-effectors to robustly track the desired trajectories. Moreover, since exactly determining the DAR system’s dynamics is impractical due to the system uncertainties, the uncertain system parameters are then proposed to be adaptively estimated by the use of the radial basis function network (RBFN). The adaptation mechanism is derived from the Lyapunov theory, which theoretically guarantees stability of the closed-loop control system. The effectiveness of the proposed RBFN-DSC approach is demonstrated by implementing the algorithm in a synthetic environment with realistic parameters, where the obtained results are highly promising.

Abstract: The processes’ modeling is an important aspect in the industry since it allows to obtain high productivity with energy and material savings. The hybrid process electrochemical discharge machining (ECDM) is subject to uncertainty and inaccuracy levels in the parameters, so a viable option to model this process is through soft computing techniques such as fuzzy logic and artificial neural networks. In this work, we present three models using fuzzy logic, backpropagation network, and radial basis function network for the prediction of the material removal rate (MRR). The gap voltage (Vg), peak current (Ip), and frequency (f) were taken as input parameters. A 3-factor full factorial design was developed with 2 levels (23), two replicas, and four central points. The model with the higher accuracy according to experimental result was radial basis function artificial neural network with 97.25% of accuracy.

Abstract: The auto-transformer rectifier unit (ATRU) is one of the most widely used avionic secondary power supplies. Timely fault identification and location of the ATRU is significant in terms of system reliability. A two-level fault diagnosis method for the ATRU using multi-source features (MSF) is proposed in this paper. Based on the topology of the ATRU, three key electrical signals are selected and analyzed to extract appropriate features for fault diagnosis. Mathematic expressions and simulation results of the feature signals under different fault modes are presented in the paper. Therefore, a unique MSF system is developed and a two-level fault diagnosis method based on radial basis function network groups is proposed. On the first level, the overall fault set is classified into three subsets and then on the second level, three radial basis function neural networks are constructed and trained to realize accurate fault localization. To verify the diagnosis performance of the proposed method, several comparative tests are implemented on a 12-pulse ATRU system, which shows that this method has a lower computational cost, better diagnostic accuracy and increased stability when compared with alternative methods.

Abstract: It has been proposed that pre-stressed concrete bridges improve load performance by inducing axial pre-stress using pre-stress tendons. However, the tensile force of the pre-stress tendons could no...

Abstract: In this paper, we propose using particle swarm optimization (PSO) which can improve weighted k-nearest neighbors (PWKNN) to diagnose the failure of a wind power system. PWKNN adjusts weight to correctly reflect the importance of features and uses the distance judgment strategy to figure out the identical probability of multi-label classification. The PSO optimizes the weight and parameter k of PWKNN. This testing is based on four classified conditions of the 300 W wind generator which include healthy, loss of lubrication in the gearbox, angular misaligned rotor, and bearing fault. Current signals are used to measure the conditions. This testing tends to establish a feature database that makes up or trains classifiers through feature extraction. Not lowering the classification accuracy, the correlation coefficient of feature selection is applied to eliminate irrelevant features and to diminish the runtime of classifiers. A comparison with other traditional classifiers, i.e., backpropagation neural network (BPNN), k-nearest neighbor (k-NN), and radial basis function network (RBFN) shows that PWKNN has a higher classification accuracy. The feature selection can diminish the average features from 16 to 2.8 and can reduce the runtime by 61%. This testing can classify these four conditions accurately without being affected by noise and it can reach an accuracy of 83% in the condition of signal-to-noise ratio (SNR) is 20dB. The results show that the PWKNN approach is capable of diagnosing the failure of a wind power system.

Abstract: Deriving operating rules for multi-objective cascade reservoir systems is an important challenge in water resources management. To address, this study combines a radial basis function network with an evolutionary algorithm to propose a heuristic input variable selection (HIS) method that extracts reservoir operating rules based on feature selection. For a case study of the Hanjiang cascade reservoirs in China, we initially describe the operating rules with radial basis functions and subsequently refine them based on the HIS method. We select the most suitable input variables for each reservoir conditioned on water supply and power generation targets to derive and optimize the rules with a Pareto-archived dynamically dimensioned search algorithm. From this we can analyze input variable selection and the corresponding impact on multi-objective cascade reservoir operations. The results demonstrate that the HIS method selects the input variables accurately and the reservoir operating rules refined by the method could increase water supply by up to 6.6% and power generation by up to 1.2%. The most suitable input variables for reservoir operation vary depending on reservoir objective, however the HIS method appears effective at selecting the appropriate input variables for individual reservoirs in a cascade system.

Abstract: The paper presents a new approach to predicting the 24-h electricity power demand in the Polish Power System (PPS, or Krajowy System Elektroenergetyczny—KSE) using the deep learning approach. The prediction system uses a deep multilayer autoencoder to generate diagnostic features and an ensemble of two neural networks: multilayer perceptron and radial basis function network and support vector machine in regression model, for final 24-h forecast one-week advance. The period of the data that is the subject of the experiments is 2014–2019, which has been divided into two parts: Learning data (2014–2018), and test data (2019). The numerical experiments have shown the advantage of deep learning over classical approaches of neural networks for the problem of power demand prediction.

Abstract: A technique for estimating a plasma drift velocity distribution in the ionosphere is presented. This technique is based on a framework for representing a global vector field on a sphere by using a set of localized basis functions which is newly derived as a variant of the spherical elementary current system (SECS). A vector field on a sphere can be divided into its divergence-free (DF) component and curl-free (CF) component. The DF and CF components can then be represented by weighted sums of the DF and CF vector-valued basis functions, respectively. While the SECS basis functions have a singular point, the new basis functions do not diverge over a sphere. This property of the new basis function allows us to achieve robust prediction of the drift velocity at any point in the ionosphere. Assuming that the ionospheric plasma drift velocity has no divergence, its distribution can be represented by a weighted sum of the DF basis functions. The proposed technique estimates the ionospheric plasma drift velocity distribution from the SuperDARN data by using the DF basis functions. Since there are some wide gaps in the spatial coverage of the SuperDARN, an empirical convection model is combined with the framework based on the new basis functions. It is demonstrated that the proposed technique is useful for the estimation and modeling of the ionospheric plasma velocity distribution.

Abstract: This paper compares the classification ability of a few efficient classifiers in recognizing human speech emotions in terms of accuracy, computation time, and feature dimension. Both the static and dynamic mel-frequency cepstral coefficients (MFCC) are derived in the wavelet domain and are combined to form a suitable identification system model. Three popular neural network (NN) models such as the Gaussian mixture model (GMM), radial basis function network (RBFN), and the probabilistic neural network (PNN) have been put to test the reliability of these derived feature sets. The PNN classifier has shown to outperform both the RBFN and the GMM with low feature dimension, whereas the GMM shows an improved result for large feature dimensions. The combination of wavelet-based MFCCs and their dynamics remains more discriminative in classifying speech emotions as compared to either the MFCCs or wavelets acted alone.

Abstract: The Artificial Intelligence (AI) is revolutionizing extensively in various industrial fields. The robustness of AI comes from utilization of information processing in solving complex real world problems. Contrary to other types of artificial intelligence, the Artificial Neural Networks (ANN) can monitor any industrial process, inspired by the functionality of the human brain. This paper is devoted to the diagnosis of induction machine by using the artificial neural network based on the stator current analysis as input features. The current work aims to compare the effectiveness of both types ANN classifiers: the Radial Basis Function Network (RBF) and Probabilistic Neural Network (PNN) in asynchronous machine faults (rotor and bearing faults) detection and severity evaluation. We've proved that RBF networks are better suited for assessing the severity of defects while the PNN gives better results when differentiating between rotor and bearing defects. The results presented in this work are confirmed experimentally.

Abstract: The presence of noise causes disturbance in the signal received by the sensor which is installed on the transformer. A nonlinear signal processing is required to effectively extract the signal. In this article, the radial basis function network (RBFN) was used to extract and eliminate the unwanted signal in case of the presence of a partial discharge source in the transformer. Elimination of the signal is due to the potential of the RBFN to approximate the nonlinear functions. The transformer, partial discharge, and noise sources are simulated in the CST software environment. Studies showed that the RBFN can successfully extract the unwanted source when it is integrated with the partial discharge signal.

Abstract: The varying system parameters, end effector payload and environmental uncertainties are quite natural in real-world robotics applications. Therefore in order to adapt to the changing control environment and improving robustness of the controller due to an uncertain system model and dynamic uncertainties adaptive control methods are developed. This paper presents an adaptive neural backstepping control for an uncertain robot manipulator with dynamic disturbances. The dynamics of an n-link uncertain robot manipulator with dynamic disturbances is expressed as a class of 3nth order nonlinear multi-input multioutput (MIMO) system using a nonlinear disturbance observer-based model. The uncertain plant and disturbances dynamics are approximated using Radial Basis Function Network (RBFN) to derive the control law. The proposed controller for each link has a simple structure with a single unknown parameter. The update law for this unknown parameter has been obtained using Lyapunov stability. It is shown that the proposed controller is able to ensure the semi-global uniformly ultimately boundedness (UUB) of all signals of the resulting closed-loop system and the actual response eventually reaches a bounded neighbourhood of the desired response. Simulation results demonstrate the feasibility of the proposed technique. The tracking performance of the proposed controller is validated experimentally on a four degrees-of-freedom (4 DOF) Barrett Whole Arm Manipulator while performing dynamic ball hitting experiments.

Abstract: The curse of dimensionality has long been a hurdle in the analysis of complex data in areas such as computational biology, ecology and econometrics. In this work, we present a forecasting algorithm that exploits the dimensionality of data in a nonparametric autoregressive framework. The main idea is that the dynamics of a chaotic dynamical system consisting of multiple time-series can be reconstructed using a combination of different variables. This nonlinear autoregressive algorithm uses multivariate attractors reconstructed as the inputs of a neural network to predict the future. We show that our approach, attractor ranked radial basis function network (AR-RBFN) provides a better forecast than that obtained using other model-free approaches as well as univariate and multivariate autoregressive models using radial basis function networks. We demonstrate this for simulated ecosystem models and a mesocosm experiment. By taking advantage of dimensionality, we show that AR-RBFN overcomes the shortcomings of noisy and short time-series data.

Abstract: Phishing attacks have become the most common cyber-security threat faced by the online users, in which the online credentials of the user are vulnerable to commit the financial crimes. The loss of money and loss of reputation of organization are the notable issues of this attack. Although several anti-phishing strategies have been proposed in the literature, this crucial issue still solicits the detection technique that favors high detection accuracy and low false alarm in the online community. Hence, in this work the Radial Basis Function (RBF) network with its enhanced hyper parameters is proposed for predicting the phishing websites. The proposed detection model utilizes the unsupervised learning for estimating the RBF kernels as well as the spread constant and obtained the categorical RBF for detecting the phishing websites. The proposed approach was tested on benchmark phishing datasets and the performance was compared with the existing neural network classifiers to prove its effectiveness.

Abstract: Background: The obstructive sleep apnea (OSA) detection has become a hot research topic because of the high risk of this disease. In this paper, we tested some powerful and low computational signal processing techniques for this task and compared their results with the recent achievements in OSA detection. Methods: The Dual-tree complex wavelet transform (DT-CWT) is used in this paper to extract feature coefficients. From these coefficients, eight non-linear features are extracted and then reduced by the Multi-cluster feature selection (MCFS) algorithm. The remaining features are applied to the hybrid “K-means, RLS” RBF network which is a low computational rival for the Support vector machine (SVM) networks family. Results and Conclusion: The results showed suitable OSA detection percentage near 96% with a reduced complexity of nearly one third of the previously presented SVM based methods.

Abstract: Learning from an imbalanced data set presents a tricky problem in which traditional learning algorithms perform poorly. Traditional classifiers usually aim to optimize the overall accuracy without ...

Abstract: In this paper, we investigate feasibility of employing Radial Basis Function (RBF) network into non-coherent detection process, for detection of targets embedded in sea clutter of unknown statistics. We particularly have in mind Croatian part of Adriatic Sea, the local sea whose clutter statistic properties are not available in open literature. Performance of the detection process employing proposed RBF network is tested with simulated clutter samples based on real sea clutter data. These data were collected under sea state conditions that represent two thirds of the total wave heights in Adriatic and are chosen to represent unknown clutter statistics due to the fact that no single probability density function equally well fits amplitude distribution of the range bins under test. It is demonstrated that, compared to the traditional [zlog(z)] method, RBF network with just four components and lognormal basis function, yields operating characteristics that better match designed ones.