Search

Computational intelligence (CI)-based methods offer a practical approach to overcoming the significant challenges posed by analytical and enumeration optimization methods when dealing with complex real-world problems. However, a notable drawback of these algorithms is the need for time-consuming and computationally demanding fine-tuning procedures to achieve optimal performance. This paper proposes a novel parameterless auto-tuning meta-heuristic architecture called the self-tuning multi-layer (STML). The fundamental concept behind this architecture involves a multi-layer structure where the inner layer optimizes the main problem. In contrast, the outer layer utilizes information obtained during the search to fine-tune the performance of the inner layer. This feature eliminates manual fine-tuning, as it can autonomously handle this task. A series of mathematical and benchmark problems were employed to demonstrate the computational prowess of the STML. The results indicate its superiority over other meta-heuristic algorithms. Additionally, the STML showcases robustness, as evidenced by the numerical proximity of results obtained from different independent runs on these benchmark problems. [Display omitted] • Self-tuning multi-layer (STML) autonomously carries out the optimization. • STML architecture eliminates manual parameter calibration. • STML removes the need for initial guesses on parameter impact for optimal results. • Preliminary results highlight STML's robustness and computational efficiency. [ABSTRACT FROM AUTHOR]

European Union has introduced the European Trading System (ETS) as a tool for developing and implementing international treaties related to climate changes and to identify the most cost-effective methods for reducing greenhouse gas emissions, in particular carbon dioxide (CO 2 ), which is the most substantial. Companies producing carbon emissions must effectively manage associated costs by buying or selling carbon emission futures. Viewed from this perspective, this paper provides a model for managing the risk by buying and selling carbon emission futures by implementing techniques that leverage computational intelligence. Three computational intelligence techniques are proposed to provide accurate and timely forecasts for changes in the price of carbon: a novel hybrid neuro-fuzzy controller that forms a closed-loop feedback mechanism called PATSOS; an artificial neural network (ANN) based system; an adaptive neuro-fuzzy inference system (ANFIS). Results are based on 1074 daily carbon price observations collected to comprise a useful time-series dataset and for evaluation of the proposed techniques. The extra-sample performance of the proposed techniques is calculated. Analysis results are compared with those produced by other models. Comparison studies reveal that PATSOS is the most accurate and promising methodology for predicting the price of carbon. It is stated that this paper registers a first attempt to apply a hybrid neuro-fuzzy controller to forecasting carbon prices. [ABSTRACT FROM AUTHOR]

Identity verification based on authenticity assessment of a handwritten signature is an important issue in biometrics. There are many effective methods for signature verification taking into account dynamics of a signing process. Methods based on partitioning take a very important place among them. In this paper we propose a new approach to signature partitioning. Its most important feature is the possibility of selecting and processing of hybrid partitions in order to increase a precision of the test signature analysis. Partitions are formed by a combination of vertical and horizontal sections of the signature. Vertical sections correspond to the initial, middle, and final time moments of the signing process. In turn, horizontal sections correspond to the signature areas associated with high and low pen velocity and high and low pen pressure on the surface of a graphics tablet. Our previous research on vertical and horizontal sections of the dynamic signature (created independently) led us to develop the algorithm presented in this paper. Selection of sections, among others, allows us to define the stability of the signing process in the partitions, promoting signature areas of greater stability (and vice versa). In the test of the proposed method two databases were used: public MCYT-100 and paid BioSecure. [ABSTRACT FROM AUTHOR]

Breast cancer is the most common disease, which is the leading cause of cancer deaths among women. This deadly disease is curable, if identified at the initial stage. In this paper, we propose a scheme for predicting the real stage of breast cancer by retrieving the mammogram images from the past cases that are similar to the query image. First, Gabor and energy features are extracted to differentiate abnormal tissue textures from normal tissue texture. In the second step, back propagation neural network (BPNN) algorithm is used to detect the tumor. Next, pattern is extracted from the query image and the real stage of breast cancer is identified using the depth of the tumor. In this paper, Euclidean distance metric and Mahalanobis distance metric are used to compute the pattern similarity between the images for retrieval. In the same way, pattern base images also retrieved. The tumor found in these retrieved images shows the same stage of breast cancer related to the query image. Performance of Euclidean and Mahalanobis distance metric is compared using precision recall measures. The proposed approach achieved 87% classification rate. [ABSTRACT FROM AUTHOR]

Abstract: Granular computing is a computational paradigm that mimics human cognition in terms of grouping similar information together. Compatibility operators such as cardinality, orientation, density, and multidimensional length act on both in raw data and information granules which are formed from raw data providing a framework for human-like information processing where information granulation is intrinsic. Granular computing, as a computational concept, is not new, however it is only relatively recent when this concept has been formalised computationally via the use of Computational Intelligence methods such as Fuzzy Logic and Rough Sets. Neutrosophy is a unifying field in logics that extents the concept of fuzzy sets into a three-valued logic that uses an indeterminacy value, and it is the basis of neutrosophic logic, neutrosophic probability, neutrosophic statistics and interval valued neutrosophic theory. In this paper we present a new framework for creating Granular Computing Neural-Fuzzy modelling structures via the use of Neutrosophic Logic to address the issue of uncertainty during the data granulation process. The theoretical and computational aspects of the approach are presented and discussed in this paper, as well as a case study using real industrial data. The case study under investigation is the predictive modelling of the Charpy Toughness of heat-treated steel; a process that exhibits very high uncertainty in the measurements due to the thermomechanical complexity of the Charpy test itself. The results show that the proposed approach leads to more meaningful and simpler granular models, with a better generalisation performance as compared to other recent modelling attempts on the same data set. [Copyright &y& Elsevier]

Abstract This paper proposed a new method for detecting islanding of distributed generation (DG), using Multi-gene Genetic Programming (MGP). Islanding has been a serious concern among power distribution utilities and distributed generation owners, because it poses risks to the safety of utilities' workers and consumers, and can cause damage to power distribution systems' equipment. Therefore, a DG must be disconnected as soon as an islanding is detected. In addition, an islanding detection method must have high degree of dependability to correctly discriminate islanding from other events, such as load switching, in order to avoid unnecessary disconnection of the distributed generator. In this context, the novelty of the proposed method is that the MGP is capable of obtaining a set of mathematical and logic functions employed to detect and classify islanding correctly. This is a new approach among the computational intelligent methods proposed for DG islanding detection. The main idea was to use local voltage measurements as input of the method, eliminating the need of complex and expensive communication infrastructure. The method has been trained with several islanding and non-islanding cases, by using a power distribution system comprising five concentrated loads, a synchronous distributed generator and a wind power plant. The results showed that the proposed method was successful in differentiating the islanding events from other disturbances, revealing its great potential to be applied in anti-islanding protection schemes for distributed generation. Highlights • Multi-gene genetic programming (MGP) is used to classify islanding of DG. • As far as we know, MGP has not been used yet for islanding detection. • Proposed method has superior performance than an artificial neural ntowrk method. • The proposed method has high potential for DG islanding detection purposes. • Proposed method has better performance than traditional islanding detection methods. [ABSTRACT FROM AUTHOR]

Abstract: As complete prevention of computer attacks is not possible, intrusion detection systems (IDSs) play a very important role in minimizing the damage caused by different computer attacks. There are two intrusion detection methods: namely misuse- and anomaly-based. A collaborative, intelligent intrusion detection system (CIIDS) is proposed to include both methods, since it is concluded from recent research that the performance of an individual detection engine is rarely satisfactory. In particular, two main challenges in current collaborative intrusion detection systems (CIDSs) research are highlighted and reviewed: CIDSs system architectures and alert correlation algorithms. Different CIDSs system, architectures are explained and compared. The use of CIDSs together with other multiple security systems raise certain issues and challenges in, alert correlation. Several different techniques for alert correlation are discussed. The focus will be on correlation of CIIDS alerts. Computational, Intelligence approaches, together with their applications on IDSs, are reviewed. Methods in soft computing collectively provide understandable, and autonomous solutions to IDS problems. At the end of the review, the paper suggests fuzzy logic, soft computing and other AI techniques, to be exploited to reduce the rate of false alarms while keeping the detection rate high. In conclusion, the paper highlights opportunities for an integrated solution to large-scale CIIDS. [Copyright &y& Elsevier]

Abstract: In several application areas like banking, insurance, medicine, education, business, to name just a few, there are huge, sometimes unstructured data collections, and there is a need to convert data into information. On the other hand, decision makers in industry are usually not statisticians, mathematicians, systems engineers, or AI experts. So it is important, to organize the results of the data analysis process and to present it in a form that can be easily interpreted by non-experts. One of the potential drawbacks affecting the application of computational intelligence (CI) methods in general to the analysis of data is the often limited interpretability of the results they yield. One way to overcome interpretability limitations is by explaining the operation of CI models using rule extraction methods. The interpretability of the model results should be greatly improved by their description in terms of reasonably simple and actionable rules that decision makers could rely on. In fact, rule extraction should provide whom the final responsibility for taking decisions rests, with an explanation about how a CI or related computer-based method has reached its decision. This paper describes a novel rule-extraction algorithm based on fuzzy logic, name LR-FIR (linguistic rules in FIR), that starts from the fuzzy inductive reasoning (FIR) methodology. FIR is able to obtain good qualitative relationships between the variables that compose the system and to predict the future behaviour of that system. The proposed algorithm (LR-FIR) is able to derive linguistic rules from a FIR model. The LR-FIR functioning is similar to those used in Boolean algebra. However the premises and consequences of rules are not necessarily binary in nature, hence the algorithm must be able to deal with multi-valued logic, and accept partial do-not-care conditions. Due to the fact that LR-FIR was developed within the FIR methodology, the obtained rules could be considered as predictive rules and deal naturally with the uncertainty captured in the FIR models. LR-FIR, in this paper, was evaluated using five data-sets from different domains: e-learning, global change temperature, brain tumour diagnosis, and two of the most used classical UCI data-sets: IRIS and Pima Indian Diabetes. The rules extracted by LR-FIR capture the main behaviour of each application, from the domain experts’ point of view, demonstrating in this sense, the efficiency of the proposed algorithm. [Copyright &y& Elsevier]

Abstract Tool path planning is a key to ensure high machining quality and productivity in 5-axis flank milling of ruled surfaces. Previous studies have shown that optimization-driven tool path planning can effectively reduce the geometrical errors on the finished surface. However, to solve the corresponding optimization problem is a challenging task involving a large number of decision variables. This paper proposes a novel approach to generating optimized tool path for 5-axis flank finishing cut based on a geometric decomposition strategy and multi-population harmony search algorithm. The proposed approach geometrically divides the surface to be machined into a number of segments. The tool paths on those sub-surfaces are independently optimized by the multi-population harmony search algorithm. Individual tool paths are then combined together to form a complete one. The test results of representative surfaces show that the proposed approach produces higher machining precision with less computational time than compared previous methods. And the computational time is further reduced by Message passing interface based parallel computing techniques. A detailed analysis is conducted to characterize how the number of divisions affects the optimization results. And the proposed approach also shows good scalability with the increasing number of cutter locations. Graphical abstract Highlights • A new tool path planning method for 5-axis flank milling is proposed. • It geometrically divides the surface into a number of segments. • The multi-population harmony search is proposed to optimize tool path on each sub-surface independently. • The effectiveness of the proposed approach is demonstrated on eight representative surfaces with different characteristics. [ABSTRACT FROM AUTHOR]

Light weight composite materials (LWCM) have gained tremendous attention, thanks to their low cost, eco-friendly nature, biodegradability, life-cycle superiority, noble mechanical properties for polymer composites and excellent electrical properties for carbon composites. LWCM are widely used in various engineering fields including aerospace, automobile, civil and architecture. The diversified applications of LWCM demand an extensive investigation of their mechanical and structural behavior under elevated conditions. Therefore, modeling of LWCM for mechanical and structural properties is a challenging task. Nowadays, computational intelligence (CI) methods are ones of the immensely growing topics in materials data science including materials imaging, feature identification, process and quality control, prediction, uncertainty quantification and design optimization. CI has not only been emerged as a revolutionary way to dig out material-related information from various data sources to serve next generation composite materials with unprecedented properties but also as an active surrogate model for conventional experimental methods. CI techniques are able to increase the quality of LWCM by optimizing the characteristics of individual constituent, their orientation, volume fractions, laminas and thickness using an adequate model. The process of designing, controlling and optimizing LWCM with distinguished properties has been redefined by various CI methods. In this paper, a high-level overview of CI methods has been introduced from the beginning of raw material selection to the final output performance of LWCM, discusses the strength, challenges and limitations of CI methods for LWCM, and proposes solutions and future directions. • A new approach is discussed regarding artificial intelligence (AI). • Composite materials and their modeling with AI are discussed. • New algorithms are introduced for fiber reinforced composites. • Golden eagle, ANN, crow search algorithms are elaborated. [ABSTRACT FROM AUTHOR]

The development of computational intelligence based approaches for the compression of graphs is an under-explored area of research. Further, compression of weighted graphs is significantly more complicated than compression of unweighted graphs. In this paper a multi-objective approach using NSGA-II is applied to the problem of weighted graph compression. The approach is designed to find a balance between the level of compression and the distortion created by the compression. Distortion is measured using two fitness functions that each evaluate changes both in graph structure and in edge weights. The methodology is applied to three weighted contact networks with differing characteristics. It was found that the multi-objective approach is useful in identifying suitable compression ratios based upon defined levels of acceptable distortion, with a single-objective genetic algorithm then applied to focus on this target compression ratio to further reduce distortion. • A multi-objective approach is applied to the problem of weighted graph compression. • The approach helps find a balance between the levels of compression and distortion. • Two fitness functions were applied to three weighted contact networks. • One of the fitness functions enables easier selection of a target compression ratio. [ABSTRACT FROM AUTHOR]

Highlights • Paper describes a novel CABC algorithm formed by fusing the concepts of statistical covariance matrices with Artificial Bee Colony algorithm. • The concepts of co-variance are detailed with example iterations. • The performance of algorithm is tested on COCO benchmarks. • The performance is exhaustively analyzed and compared with state-of-the-art algorithms which are based on co-variance matrices. • Statistical tests are done to gauge the significance in adept performance. Abstract Artificial Bee Colony (ABC) is one of the proficient and largely used optimization technique, inspired by the food search behavior of honey bees. This article presents a novel Co-variance guided Artificial Bee Colony (CABC) algorithm which is a unification of ABC and statistical co-variance. The co-variance matrix of data sets is a good approximation of the Hessian and acts as a source of gradient information in meta-heuristic optimization. In this article we have demonstrated how to use covariance information about the “population of candidate solutions” for speeding up the underlying ABC optimization technique. The COCO (COmparing Continuous Optimisers), Black Box Optimization Bench-marking 2015 (BBOB) test bed with 24 benchmarks has been used to evaluate the performance of CABC algorithm. It is observed from experimental as well as statistical results that CABC has a consistent better performance when compared with other state-of-the-art evolutionary algorithms. [ABSTRACT FROM AUTHOR]

Optimization problems of modern day power system are very challenging to resolve because of its design complexity, wide geographical dispersion and influence from many unpredictable factors. For that reason, it is essential to apply most effective optimization techniques by taking full benefits of simplified formulation and execution of a particular problem. This study presents a summary of significant hybrid bio-inspired computational intelligence (CI) techniques utilized for power system optimization. Authors have reviewed an extensive range of hybrid CI techniques and examined the motivations behind their improvements. Various applications of hybrid bio-inspired CI algorithms have been highlighted in this paper. In addition, few drawbacks regarding the hybrid CI algorithms are explained. Current trends in CI techniques from the past researches have also been discussed in the domain of power system optimization. Lastly, some future research directions are suggested for further advancement of hybrid techniques. [ABSTRACT FROM AUTHOR]

This paper proposes an interval-valued multi-criteria method for learning preferences and attitudes, identifying priorities with maximal robustness for decision support. The method is based on the notion of weighted overlap dominance, formalized by means of aggregation operators and interval-valued fuzzy sets. The procedure handles uncertainty by estimating the likelihood of dominance among pairs of alternatives, inducing an attitude-based system of dominance and indifference relations. This system allows conflicting situations of indifference/dependency to arise, which need to be resolved for properly identifying preferences under any attitude. In order to do so, relevance functions are examined over the whole system of relations, obtaining a weak preference order together with its associated attitude and robustness index. As a result, the proposed method allows learning preferences and attitudes, identifying the solutions with maximal robustness for intelligent decision support. [ABSTRACT FROM AUTHOR]

Evolutionary optimization algorithms, a meta-heuristic approach, often encounter considerable challenges in many-objective optimization problems (MaOPs). The Pareto-based dominance loses its effectiveness in MaOPs, which are defined as having more than three objectives. Therefore, a more valid selection method is proposed to balance convergence and distribution. This paper proposes an algorithm using rotary grid technology to solve MaOPs (denoted by RGridEA). The algorithm uses the rotating grid to partition the objective space. Instead of using the Pareto non-dominated sorting strategy to layer the population a novel stratified method is used to enhance convergence effectively and make use of the grid to improve distribution and uniformity. Finally, with the other seven algorithm was tested on the test function DTLZ series analysis, confirming RGridEA is effective in resolving MaOPs. [ABSTRACT FROM AUTHOR]

This paper proposes a novel modified cuckoo search algorithm (NMCSA) to solve optimal placement of actuators (OPA) for active vibration control. The purpose of OPA is to minimize control spillover effect and maximize the control force applied to the desired modes. To achieve this objective, NMCSA first employs speed factor (SFR) and aggregation factor (AFR) for recording and analyzing the current and history information of nests. Secondly, SFR and AFR are mapped to suitable space by scale conversion factors (SCF). Thus, the NMCSA based on SCF can give adaptively actions on the step size α and discovery probability p a to balance exploration and exploitation. The performance of NMCSA is confirmed by some well-known benchmark functions. Subsequently, the NMCSA is applied to solve OPA and compared with several state-of-the-art algorithms in the literature, the statistical results demonstrate that the proposed algorithm has a higher convergence speed and better search ability. [ABSTRACT FROM AUTHOR]

Optimal power flow (OPF) refers to the problem of optimizing the operating decisions such as electric power generation in power systems, which are always subjected to dynamic factors like bus loads. Conventionally, OPF in dynamic environments has been solved by static-oriented optimization methods based on the prediction of the dynamic factors. However, as the dynamics of modern power systems become more and more complex and difficult to predict, research interest of intelligent methods that track the optimal decisions of OPF has been grown recently. Devoted to this objective, a learning enhanced differential evolution (LEDE) is proposed in this paper. LEDE incorporates the idea of nearest-neighbor rule from the field of machine learning, with which decisions of the previous environments are retrieved continually to replace the newly generated individuals of differential evolution. A so-called elitism stochastic ranking strategy is also proposed, used in LEDE to handle constraints of OPF. Experiments are conducted on the dynamic IEEE 30-bus system and IEEE 118-bus system, and the results show the efficiency of LEDE in comparison with other algorithms. [ABSTRACT FROM AUTHOR]

Consensus reaching processes (CRPs) in group decision making (GDM) attempt to reach a mutual agreement among a group of decision makers before making a common decision. Different consensus models have been proposed by different authors in the literature to facilitate CRPs. Classical CRP models focus on achieving an agreement on GDM problems in which few decision makers participate. However, nowadays, societal and technological trends that demand the management of larger scale of decision makers add new requirements to the solution of consensus-based GDM problems. This paper presents a comparative study of different classical CRPs applied to large-scale GDM in order to analyze their performance and find out which are the main challenges that these processes face in large-scale GDM. Such analyses will be developed in a java-based framework (AFRYCA 2.0) simulating different scenarios in large scale GDM. [ABSTRACT FROM AUTHOR]

Fuzzy classification rules are more interpretable and cope better with pervasive uncertainty and vagueness with respect to crisp rules. Because of this fact, fuzzy classification rules are extensively used in classification and decision support systems for disease diagnosis. But, most of the rule mining techniques failed to generate accurate and comprehensive fuzzy rules. This paper presents a hybrid decision support system based on Rough Set Theory (RST) and Bat optimization Algorithm (BA) called RST-BatMiner. It consists of two stages. In the first stage, redundant features have been removed from the data set through RST-based QUICK-REDUCT approach. In the second stage, for each class BA is invoked to generate fuzzy rules by minimizing proposed fitness function. Further, an Ada-Boosting technique is applied to the rules generated by BA to increase the accuracy rate of generated fuzzy rules. Moreover, to generate comprehensive fuzzy rules, a new ≠ (not equal) operator along with = (equal) operator is introduced into BA encoding scheme. The proposed RST-BatMiner builds consolidated fuzzy ruleset by learning the rules associated with each class separately. The proposed RST-BatMiner is experimented on six bench-mark datasets namely Pima Indians Diabetes, Wisconsin Breast Cancer, Cleveland Heart disease, iris, wine and glass, in order to validate its generalization capability. These experimental results show that except for wine dataset the proposed RST-BatMiner yields high accuracy and comprehensible ruleset when compared to other state-of-the-art bio-inspired based fuzzy rule miners and Fuzzy Rule Based Classification Systems (FRBCS) in the literature. In the case of wine dataset, the proposed RST-BatMiner yields second highest accuracy along with comprehensible ruleset. [ABSTRACT FROM AUTHOR]

R2 indicator based multi-objective evolutionary algorithms (R2-MOEAs) have achieved promising performance on traditional multi-objective optimization problems (MOPs) with two and three objectives, but still cannot well handle many-objective optimization problems (MaOPs) with more than three objectives. To address this issue, this paper proposes a two-stage R2 indicator based evolutionary algorithm (TS-R2EA) for many-objective optimization. In the proposed TS-R2EA, we first adopt an R2 indicator based achievement scalarizing function for the primary selection. In addition, by taking advantage of the reference vector guided objective space partition approach in diversity management for many-objective optimization, the secondary selection strategy is further applied. Such a two-stage selection strategy is expected to achieve a balance between convergence and diversity. Extensive experiments are conducted on a variety of benchmark test problems, and the experimental results demonstrate that the proposed algorithm has competitive performance in comparison with several tailored algorithms for many-objective optimization. [ABSTRACT FROM AUTHOR]

Within the software testing domain, determining the order in which classes are tested in integration testing is an important problem called class integration test order identification (CITO). This problem is useful in integration testing , as it contributes to reducing the time needed for testing a software system and refers to the process of identifying an optimal order in which the application classes should be combined and tested as a group. This paper introduces a novel approach based on reinforcement learning for class integration test order optimization in the context of integration testing . The experimental evaluation is performed on four synthetic examples and on six existing software systems often used in the literature for this problem. The results obtained are analyzed and compared to similar related work from the literature, highlighting the potential of the current proposal. The proposed reinforcement learning based approach outperforms most methods existing in the software engineering literature for optimizing the test order for class-based integration. Moreover, it is general, and can be easily extended for optimizing the order in which software components should be tested during component integration testing of a software system. [ABSTRACT FROM AUTHOR]

A parallel Jaya algorithm implemented on the graphics processing unit (GPU-Jaya) is proposed to estimate parameters of the Li-ion battery model in this paper. Similar to the generic Jaya algorithm (G-Jaya), the GPU-Jaya is free of tuning algorithm-specific parameters. Compared with the G-Jaya algorithm, three main procedures of the GPU-Jaya, the solution update, fitness value computation, and the best/worst solution selection are all computed in parallel on GPU via a compute unified device architecture (CUDA). Two types of memories of CUDA, the global memory and the shared memory are utilized in the execution. The effectiveness of the proposed GPU-Jaya algorithm in estimating model parameters of two Li-ion batteries is validated via real experiments while its high efficiency is demonstrated by comparing with the G-Jaya and other considered benchmarking algorithms. The experimental results reflect that the GPU-Jaya algorithm can accurately estimate battery model parameters while tremendously reduce the execution time using both entry-level and professional GPUs. [ABSTRACT FROM AUTHOR]

With the ongoing development of Earth observation techniques, huge amounts of remote sensing images with a high spectral-spatial-temporal resolution are now available, and have been successfully applied in a variety of fields. In the process, they bring about great challenges, such as high-dimensional datasets (the high spatial resolution and hyperspectral features), complex data structures (nonlinear and overlapping distributions), and the nonlinear optimization problem (high computational complexity). Computational intelligence techniques, which are inspired by biological systems, can provide possible solutions to the above-mentioned problems. In this paper, we provide an overview of the application of computational intelligence technologies in optical remote sensing image processing, including: 1) feature representation and selection; 2) classification and clustering; and 3) change detection. Subsequently, the core potentials of computational intelligence for optical remote sensing image processing are delineated and discussed. [ABSTRACT FROM AUTHOR]

Financial time series forecasting is undoubtedly the top choice of computational intelligence for finance researchers in both academia and the finance industry due to its broad implementation areas and substantial impact. Machine Learning (ML) researchers have created various models, and a vast number of studies have been published accordingly. As such, a significant number of surveys exist covering ML studies on financial time series forecasting. Lately, Deep Learning (DL) models have appeared within the field, with results that significantly outperform their traditional ML counterparts. Even though there is a growing interest in developing models for financial time series forecasting, there is a lack of review papers that solely focus on DL for finance. Hence, the motivation of this paper is to provide a comprehensive literature review of DL studies on financial time series forecasting implementation. We not only categorized the studies according to their intended forecasting implementation areas, such as index, forex, and commodity forecasting, but we also grouped them based on their DL model choices, such as Convolutional Neural Networks (CNNs), Deep Belief Networks (DBNs), and Long-Short Term Memory (LSTM). We also tried to envision the future of the field by highlighting its possible setbacks and opportunities for the benefit of interested researchers. • We reviewed all searchable articles of deep learning (DL) for financial time series forecasting. • RNN based DL models (LSTM and GRU included) are the most common. • We compared DL models according to their performances in different forecasted asset classes. • To best of our knowledge, this is the first comprehensive DL survey for financial time series forecasting. • We provided current status of DL in financial time series forecasting, also highlighted the future opportunities. [ABSTRACT FROM AUTHOR]

Sentiment analysis, also called opinion mining, is currently one of the most studied research fields which aims to analyse people's opinions. E-commerce websites allow users to share opinions about a product/service by providing textual reviews along with numerical ratings. These opinions greatly influence future consumer purchasing decisions. This paper introduces an innovative computational intelligence framework for efficiently predicting customer review ratings. The framework has been designed to deal with the dimensionality and noise which is typically apparent in large datasets containing customer reviews. The proposed framework integrates the techniques of Singular Value Decomposition (SVD) and dimensionality reduction, Fuzzy C-Means (FCM) and the Adaptive Neuro-Fuzzy Inference System (ANFIS). The performance of the proposed approach returned high accuracy and the results revealed that when large datasets are concerned, only a fraction of the data is needed for creating a system to predict the review ratings of textual reviews. Results from the experiments suggest that the proposed approach yields better prediction performance than other state-of-the-art rating predictors which are based on the conventional Artificial Neural Network, Fuzzy C-Means, and Support Vector Machine approaches. In addition, the proposed framework can be utilised for other classification and prediction tasks, and its neuro-fuzzy predictor module can be replaced by other classifiers. [ABSTRACT FROM AUTHOR]

In this paper, an intelligent adaptive tracking control system (IATCS) based on the mixed H 2 /H ∞ approach under uncertain plant parameters and external disturbances for achieving high precision performance of a two-axis motion control system is proposed. The two-axis motion control system is an X – Y table driven by two permanent-magnet linear synchronous motors (PMLSMs) servo drives. The proposed control scheme incorporates a mixed H 2 /H ∞ controller, a self-organizing recurrent fuzzy-wavelet-neural-network controller (SORFWNNC) and a robust controller. The combinations of these control methods would insure the stability, robustness, optimality, overcome the uncertainties, and performance properties of the two-axis motion control system. The SORFWNNC is used as the main tracking controller to adaptively estimate an unknown nonlinear dynamic function that includes the lumped parameter uncertainties, external disturbances, cross-coupled interference and frictional force. Moreover, the structure and the parameter learning phases of the SORFWNNC are performed concurrently and online. Furthermore, a robust controller is designed to deal with the uncertainties, including the approximation error, optimal parameter vectors and higher order terms in Taylor series. Besides, the mixed H 2 /H ∞ controller is designed such that the quadratic cost function is minimized and the worst case effect of the unknown nonlinear dynamic function on the tracking error must be attenuated below a desired attenuation level. The mixed H 2 /H ∞ control design has the advantage of both H 2 optimal control performance and H ∞ robust control performance. The sufficient conditions are developed for the adaptive mixed H 2 /H ∞ tracking problem in terms of a pair of coupled algebraic equations instead of coupled nonlinear differential equations. The coupled algebraic equations can be solved analytically . The online adaptive control laws are derived based on Lyapunov theorem and the mixed H 2 /H ∞ tracking performance so that the stability of the proposed IATCS can be guaranteed. Furthermore, the control algorithms are implemented in a DSP-based control computer. From the experimental results, the motions at X -axis and Y -axis are controlled separately, and the dynamic behaviors of the proposed IATCS can achieve favorable tracking performance and are robust to parameter uncertainties. [ABSTRACT FROM AUTHOR]

Optimal reactive power dispatch (ORPD) is one of the paramount issue for the researchers during the investigation of power system performance in dynamic load scenarios. In this paper, a new nature inspired computing paradigm based on fractional order comprehensive learning particle swarm optimization (FO-CLPSO) is designed and implemented for solving the reactive power dispatch problems. The objective of the study is to improve the power system efficiency by reducing line losses, enhancing bus voltage profiles and reducing the operating cost of the system for different load factors. The decision variables for the fitness evaluation are the tap changer settings, generator bus voltages, fixed capacitors and flexible AC transmission systems (FACTS). The operation, validity and scalability of the FO-CLPSO are tested on standard IEEE 30 bus and IEEE-57 bus systems. The exploitation and exploration for FO-CLPSO are further extended using different fractional orders for minimization problems in ORPD to critically analyze the performance by comparing with several state of art counterpart methodologies. The stability, consistency, robustness and reliability of FO-CLPSO for the solution of ORPD problems is also substantiated through detailed statistical analyses including the development of empirical cumulative distribution functions, probability plots, box plot illustrations and histograms both for precision and complexity metrics. • Fractional order comprehensive learning PSO is proposed for ORPD problems. • Performance investigation of FO-CLPSO under different load conditions. • The consistency and reliability assessment through statistical observations. [ABSTRACT FROM AUTHOR]

Modeling the glass-forming ability (GFA) of bulk metallic glasses (BMGs) is one of the hot issues ever since bulk metallic glasses (BMGs) are discovered. It is very useful for the development of new BMGs for various engineering applications, if GFA criterion modeled precisely. In this paper, we have proposed support vector regression (SVR), artificial neural network (ANN), general regression neural network (GRNN), and multiple linear regression (MLR) based computational intelligent (CI) techniques that model the maximum section thickness ( D max ) parameter for glass forming alloys. For this study, a reasonable large number of BMGs alloys are collected from the current literature of material science. CI models are developed using three thermal characteristics of glass forming alloys i.e., glass transition temperature ( T g ), the onset crystallization temperature ( T x ), and liquidus temperature ( T l ). The R 2 -values of GRNN, SVR, ANN, and MLR models are computed to be 0.5779, 0.5606, 0.4879, and 0.2611 for 349 BMGs alloys, respectively. We have investigated that GRNN model is performing better than SVR, ANN, and MLR models. The performance of proposed models is compared to the existing physical modeling and statistical modeling based techniques. In this study, we have investigated that proposed CI approaches are more accurate in modeling the experimental D max than the conventional GFA criteria of BMGs alloys. [ABSTRACT FROM AUTHOR]

Data analysis techniques have been traditionally conceived to cope with data described in terms of numeric vectors. The reason behind this fact is that numeric vectors have a well-defined and clear geometric interpretation, which facilitates the analysis from the mathematical viewpoint. However, the state-of-the-art research on current topics of fundamental importance, such as smart grids, networks of dynamical systems, biochemical and biophysical systems, intelligent trading systems, multimedia content-based retrieval systems, and social networks analysis, deal with structured and non-conventional information characterizing the data, providing richer and hence more complex patterns to be analyzed. As a consequence, representing patterns by complex (relational) structures and defining suitable, usually non-metric, dissimilarity measures is becoming a consolidated practice in related fields. However, as the data sources become more complex, the capability of judging over the data quality (or reliability) and related interpretability issues can be seriously compromised. For this purpose, automated methods able to synthesize relevant information, and at the same time rigorously describe the uncertainty in the available datasets, are very important: information granulation is the key aspect in the analysis of complex data. In this paper, we discuss our general viewpoint on the adoption of information granulation techniques in the general context of soft computing and pattern recognition, conceived as a fundamental approach towards the challenging problem of automatic modeling of complex systems. We focus on the specific setting of processing the so-called non-geometric data, which diverges significantly from what has been done so far in the related literature. We highlight the motivations, the founding concepts, and finally we provide the high-level conceptualization of the proposed data analysis framework. [ABSTRACT FROM AUTHOR]

The tools of soft computing will aid the knowledge mining in predicting and classifying the properties of various parameters while designing the composite preforms in the manufacturing of Powder Metallurgy (P/M) Lab. In this paper, an integrated PRNET (PCA-Radial basis functional neural NET) model is proposed in different versions to select the relevant parameters for preparing composite preforms and to predict the deformation and strain hardening properties of Al–Fe composites. It reveals that the predictability of this model has been increased by 67.89% relatively from the conventional models. A new PR-filter is proposed by slightly modifying the conventional filters of RBFNN, which improves the power of PRNET even though raw data are highly non-linear, interrelated and noisy. Moreover, fixing the range of input parameters for classifying the properties of composite preforms can be automated by the Fuzzy logic. These types of models will avoid expensive experimentation and risky environment while preparing sintered composite preforms. Thus the manufacturing process of composites in P/M Lab will be simplified with minimum energy by the support of these soft-computing tools. [ABSTRACT FROM AUTHOR]

Central force optimization (CFO) is an efficient and powerful population-based intelligence algorithm for optimization problems. CFO is deterministic in nature, unlike the most widely used metaheuristics. CFO, however, is not completely free from the problems of premature convergence. One way to overcome local optimality is to utilize the multi-start strategy. By combining the respective advantages of CFO and the multi-start strategy, a multi-start central force optimization (MCFO) algorithm is proposed in this paper. The performance of the MCFO approach is evaluated on a comprehensive set of benchmark functions. The experimental results demonstrate that MCFO not only saves the computational cost, but also performs better than some state-of-the-art CFO algorithms. MCFO is also compared with representative evolutionary algorithms. The results show that MCFO is highly competitive, achieving promising performance. [ABSTRACT FROM AUTHOR]

Soft computing is an interdisciplinary area that focuses on the design of intelligent systems to process uncertain, imprecise and incomplete information. It mainly builds on fuzzy sets theory, fuzzy logic, neural computing, optimization, evolutionary algorithms, and approximate reasoning et al. Information granularity is in general regarded as a crucial design asset, which helps establish a better rapport of the resulting granular model with the system under modeling. Human centricity is an inherent property of people's view on a system, a process, a machine or a model. Information granularity can be used to reflect people's level of uncertainty and this makes its pivotal role in soft computing. Indeed, the concept of information granularity facilitates the development of theory and application of soft computing immensely. A number of papers pertaining to some recent advances in theoretical development and practical application of information granularity in soft computing are highlighted in this special issue. The main objective of this study is to collect as many as possible researches on human centricity and information granularity in the agenda of theories and applications of soft computing, review the main idea of these literatures, compare the advantages and disadvantages of their methods and try to find the relationships and relevance of these theories and applications. [ABSTRACT FROM AUTHOR]

In recent years, the use of magnetic field measurements has become relevant in several applications ranging from non-invasive structural fault detection to tracking of micro-capsules within living organisms. Magnetic measurements are, however, affected by a high noise due to a number of causes, such as interference from external objects and the Earth magnetic field. Furthermore, in many situations the magnetic fields under analysis are time-variant, for example because generated by moving objects, power lines, antennas, etc. For these reasons, a general approach for accurate real-time magnetic dipole detection is unfeasible, but specific techniques should be devised. In this paper we explore the possibility of using multiple 3-axis magnetic field sensors to estimate the position and orientation of a magnetic dipole moving within the detection area of the sensors. We propose a real-time Computational Intelligence approach, based on an innovative single particle optimization algorithm, for solving, with an average period of 2.5 ms, the inverse problem of magnetic dipole detection. Finally, we validate the proposed approach by means of an experimental setup consisting of 3 sensors and a custom graphical application showing in real-time the estimated position and orientation of the magnetic dipole. Experimental results show that the proposed approach is superior, in terms of detection error and computational time, to several state-of-the-art real-parameter optimization algorithms. [ABSTRACT FROM AUTHOR]

Stock selection is an important decision making problem. Trading strategies and rules based on fundamental and technical analysis can be used for decision making process. In this paper, we propose an intelligent stock selection method, which is called case-based reasoning (CBR). This technique uses the fundamental and technical indicators to identify the winning stocks around the earning announcements. CBR method is compared with other artificial intelligence techniques such as multi layer perceptron (MLP), decision trees (QUEST, Classification and Regression Trees, C5), generalized rule induction (GRI) and logistic regression. We show that the performance of CBR is better than the performance of other techniques in terms of classification accuracy, average return, Sharpe ratio and ideal profit. [ABSTRACT FROM AUTHOR]

In this paper we present a model based on computational intelligence and natural language generation for the automatic generation of textual summaries from numerical data series, aiming to provide insights which help users to understand the relevant information hidden in the data. Our model includes a fuzzy temporal ontology with temporal references which addresses the problem of managing imprecise temporal knowledge, which is relevant in data series. We fully describe a real use case of application in the environmental information systems field, providing linguistic descriptions about the air quality index (AQI), which is a very well-known indicator provided by all meteorological agencies worldwide. We consider two different data sources of real AQI data provided by the official Galician (NW Spain) Meteorology Agency: (i) AQI distribution in the stations of the meteorological observation network and (ii) time series which describe the state and evolution of the AQI in each meteorological station. Both application models were evaluated following the current standards and good practices of manual human expert evaluation of the Natural Language Generation field. Assessment results by two experts meteorologists were very satisfactory, which empirically confirm that the proposed textual descriptions fit this type of data and service both in content and layout. • In this work we present a natural language generation model for the automatic generation of summaries to describe numerical data series. • We developed a fuzzy temporal model to manage the imprecision of temporal knowledge. • We provide a real use case of this model in the meteorological area providing summaries from air quality index data. • A human expert evaluation of the resulting linguistic descriptions was performed. [ABSTRACT FROM AUTHOR]

Abstract: It is important to properly assess the weights of attributes when solving multi-attribute decision problems because variations in the weights often influence the rankings of the alternatives. In this paper, we propose an alternative objective weighting method to generate objective weights based on intuitionistic fuzzy (IF) entropy measures, which depend on the nature of a decision matrix in an intuitionistic fuzzy environment. Instead of the traditional fuzzy entropy measure, which is characterized by using the discriminating power to calculate the attribute weights, the proposed approach adopts the IF entropy measure, which emphasizes the credibility of the data. The IF entropy measure applied here is derived from a geometric interpretation of intuitionistic fuzzy sets and incorporates the concept of a ratio of distance measures. We implement four distance measures in the proposed approach and compare them in a computational experiment. The experimental results indicate that different IF entropy measures used in weighting methods can generate distinct objective attribute weights. In particular, when the number of attributes increases, the discrepancy between the IF entropy measures increases. [Copyright &y& Elsevier]

Abstract: This paper presents the design of walking motions of a biped robot by a computational intelligent method. The proposed method finds the parameters of the output functions that produce such motion in the closed-loop system which fulfill the requirements of stability and minimal energy control by formulating a constrained optimization problem. A set of parameters which produces periodic motions with low energy control magnitudes and its closed-loop system trajectories are shown through a numerical study case of a 3-DOF biped robot. [Copyright &y& Elsevier]

Abstract: Forecasting of future electricity demand is very important for decision making in power system operation and planning. In recent years, due to privatization and deregulation of the power industry, accurate electricity forecasting has become an important research area for efficient electricity production. This paper presents a time series approach for mid-term load forecasting (MTLF) in order to predict the daily peak load for the next month. The proposed method employs a computational intelligence scheme based on the self-organizing map (SOM) and support vector machine (SVM). According to the similarity degree of the time series load data, SOM is used as a clustering tool to cluster the training data into two subsets, using the Kohonen rule. As a novel machine learning technique, the support vector regression (SVR) is used to fit the testing data based on the clustered subsets, for predicting the daily peak load. Our proposed SOM-SVR load forecasting model is evaluated in MATLAB on the electricity load dataset provided by the Eastern Slovakian Electricity Corporation, which was used in the 2001 European Network on Intelligent Technologies (EUNITE) load forecasting competition. Power load data obtained from (i) Tenaga Nasional Berhad (TNB) for peninsular Malaysia and (ii) PJM for the eastern interconnection grid of the United States of America is used to benchmark the performance of our proposed model. Experimental results obtained indicate that our proposed SOM-SVR technique gives significantly good prediction accuracy for MTLF compared to previously researched findings using the EUNITE, Malaysian and PJM electricity load datasets. [Copyright &y& Elsevier]

Abstract: Bankruptcy prediction has been a topic of active research for business and corporate organizations since past few decades. The problem has been tackled using various models viz., Statistical, Market Based and Computational Intelligence in the past. Among Computational Intelligence models, Artificial Neural Network has become dominant modeling paradigm. In this Paper, we use a novel Soft Computing tool viz., Fuzzy Support Vector Machine (FSVM) to solve bankruptcy prediction problem. Support Vector Machine is a powerful statistical classification technique based on the idea of Structural Risk Minimization. Fuzzy Sets are capable of handling uncertainty and impreciseness in corporate data. Thus, using the advantage of Machine Learning and Fuzzy Sets prediction accuracy of whole model is enhanced. FSVM is implemented for analyzing predictors as financial ratios. A method of adapting it to default probability estimation is proposed. The test dataset comprises of 50 largest bankrupt organizations with capitalization of no less than $1 billion that filed for protection against creditors under Chapter 11 of United States Bankruptcy Code in 2001–2002 after stock marked crash of 2000. Experimental results on FSVM illustrate that it is better capable of extracting useful information from corporate data. This is followed by a comparative study of FSVM with other approaches. FSVM is effective in finding optimal feature subset and parameters. This is evident from the results thus improving prediction of bankruptcy. The choice of feature subset has positive influence on appropriate kernel parameters and vice versa which demonstrate its appreciable generalization performance than traditional bankruptcy prediction methods. Choosing appropriate value of parameter plays an important role on the performance of FSVM model. The effect of variability in prediction performance of FSVM with respect to various values of different parameters of SVM is also investigated. Finally, a comparative study of clustering power of FSVM is made with PNN on ripley and bankruptcy datasets. The results show that FSVM has superior clustering power than PNN. [Copyright &y& Elsevier]

Abstract: In the era of the Web, there is urgent need for developing systems able to personalize the online experience of Web users on the basis of their needs. Web recommendation is a promising technology that attempts to predict the interests of Web users, by providing them with information and/or services that they need without explicitly asking for them. In this paper we propose NEWER, a usage-based Web recommendation system that exploits the potential of Computational Intelligence techniques to dynamically suggest interesting pages to users according to their preferences. NEWER employs a neuro-fuzzy approach in order to determine categories of users sharing similar interests and to discover a recommendation model as a set of fuzzy rules expressing the associations between user categories and relevances of pages. The discovered model is used by a online recommendation module to determine the list of links judged relevant for users. The results obtained on both synthetic and real-world data show that NEWER is effective for recommendation, leading to a quality of the generated recommendations comparable and often significantly better than those of other approaches employed for the comparison. [Copyright &y& Elsevier]

Abstract: Mass appraisal is the systematic appraisal of groups of properties as of a given date using standardized procedures and statistical testing. Mass appraisal is commonly used to compute real estate tax. There are three traditional real estate valuation methods: the sales comparison approach, income approach, and the cost approach. Mass appraisal models are commonly based on the sales comparison approach. The ordinary least squares (OLS) linear regression is the classical method used to build models in this approach. The method is compared with computational intelligence approaches – support vector machine (SVM) regression, multilayer perceptron (MLP), and a committee of predictors in this paper. All the three predictors are used to build a weighted data-depended committee. A self-organizing map (SOM) generating clusters of value zones is used to obtain the data-dependent aggregation weights. The experimental investigations performed using data cordially provided by the Register center of Lithuania have shown very promising results. The performance of the computational intelligence-based techniques was considerably higher than that obtained using the official real estate models of the Register center. The performance of the committee using the weights based on zones obtained from the SOM was also higher than of that exploiting the real estate value zones provided by the Register center. [Copyright &y& Elsevier]

Power transformers are an indispensable equipment in power transmission and distribution systems, and failures or hidden defects in power transformers can cause operational and downtime issues in power supply, resulting in economic and resource losses. Therefore, it is highly desirable to put in place intelligent preventive maintenance measures to diagnose and evaluate the condition of power transformers. Although conventional methods have achieved success in detecting problems associated with power transformers, their adoption rate in practical environments is still far from universal. The advent of Computational Intelligence (CI) models offers useful potential to complement the existing diagnostic practices of power transformers. In this paper, we provide a review on various computational intelligence techniques for fault detection and diagnosis pertaining to preventive maintenance of power transformers. An overview of each representative CI approach is presented to facilitate researchers in selecting an appropriate method for a specific problem at hand. We carry out a broad discussion on numerous concerns and challenges that are missing from the current literature, which, nevertheless, need to be addressed seriously. We identify the research gaps in the literature, and suggest the way forward in research that will in the long run enhance power system reliability by embracing CI approaches into business operations in an effort to realize the Sustainable Development Goal (SDGs) advocated by the United Nation, primarily SDG7: Clean and Affordable Energy and SDG9: Industry, Innovation and Infrastructure. • We aim to provide a knowledge base that covers more than DGA to inform future research. • A wide range of conventional methods is presented in Section 3. • The advent of Computational Intelligence (CI) models offers useful potential to complement the existing practices, in Section 4. • Numerous concerns that are missing from the current literature, are discussed in Section 5. • The existing studies are highlighted and the research gaps are identified to guide future research. [ABSTRACT FROM AUTHOR]

Artificial Intelligence (AI) technologies are widely used to study the computational drug repositioning to find potential new uses for marketed drugs, which have a huge impact on drug development. However, due to the sparsity of datasets and the linear limitations of traditional AI or machine learning models lead to the inability of most methods to effectively obtain hidden feature of drugs and diseases To end this, Predictive Intelligence (PI) has been proposed in recent years. Hence, in this work, we propose a PI powered Attentional Stacking Matrix Factorization (PI-ASMF). Firstly, in order to extract the effective latent factor of the drugs or diseases the ASMF model superimposes the drug–disease similarity information and the drug–disease association information into the PI model, which ensures that it can learn the effective latent factor and alleviates the cold start problem to a certain extent. Subsequently, the latent factors of drugs and diseases are element-wise multiplied with their respective attentional vectors to learn the adaptive weights of each feature. Finally, the latent factors of drugs and diseases are fed into the prediction module to generate predictive values of potential drug–disease associations. The AUC and AUPR values of PI-ASMF model on two real datasets are 0.892 and 0.168, 0.913 and 0.25, respectively, which verified its superiority and validity. • The novel PI-ASMF model is proposed in this paper. • The PI-ASMF model can learn the effective drug or disease latent factor. • The PI-ASMF model can capture more complex drug–disease associations. • The results prove the effectiveness and superiority of the PI-ASMF model. [ABSTRACT FROM AUTHOR]

IoMT sensors such as wearables, moodables, ingestible sensors and trackers have the potential to provide a proactive approach to healthcare. But grouping, traversing and selectively tapping the IoMT data traffic and its immediacy makes data management & decision analysis a pressing issue. Evidently, the selection process for real-world, time-constrained health problems involves looking at multivariate time-series data generated simultaneously from various wearables resulting in data overload and accuracy issues. Computational intelligence of edge analytics can extend predictive capability by quickly turning digital biomarker data into actions for remote monitoring and trigger alarm during emergency incidents without relying on backend servers. But the pervasive generation of data streams from IoMT levies significant issues in data visualization and exploratory data analysis. This paper presents a genetically optimized Fuzzy C-means data clustering technique for affective state recognition on the edge. Clustering segregates the biomarker data in chunks and generates a summarized data for each subject which is then genetically optimized to avoid stagnation in local optima. A multi-level convolution neural network is finally used to classify the affective states into the baseline, stress and amusement categories. The model is evaluated on the publicly available WESAD dataset and compares favorably to state-of-the-art with less time complexity. It demonstrates the use of data clustering technique for numerosity reduction of real-time data streams in intelligent edge analytics which facilitates fast analysis of affective state of the user. • Continuous IoMT-based data analysis showcasing computational intelligence on the edge. • Extending predictive capabilities for turning digital biomarker data into actions. • Optimized Fuzzy C-means data clustering for data summarization. • Deep hierarchical model for affective state detection on edge using optimized data. • Validated on publicly available WESAD dataset IoMT biomarker data • Compares favorably to state-of-the-art with less time complexity. [ABSTRACT FROM AUTHOR]

Smart Internet of Things (smart IoT) have emerged as a transformative computing paradigm recently. This new approach has made great contributions in the area of cyber–physical–social systems by employing various computational intelligence techniques like deep learning, for analyzing data, especially heterogeneous data from sensing and wireless communication. As a representative example of deep learning, deep residual networks have achieved excellent performance for big data feature learning since they can avoid gradient vanishing issues in deep learning models effectively. Unfortunately, they could not learn features for heterogeneous data, especially multi-modal data, in smart IoT. This paper proposes a deep residual computation model by generalizing the deep residual network in the tensor space. Especially, each multi-modal data object is represented as a tensor, while all hidden layers are also represented as tensors. Furthermore, we propose a tensor back-propagation algorithm to train the parameters of the deep residual computation model. Finally, we conduct extensive experiments to evaluate the presented deep residual model by comparing with the existing models such as multi-modal deep learning models, 3D deep residual models, deep computation models, and deep convolutional computation models. Results show that the proposed model produces more accurate classification results than other models for heterogeneous data feature learning in cyber–physical–social systems. • We present a deep residual computation model for heterogeneous data feature learning. • We propose a tensor back-propagation algorithm to train the parameters of the deep residual computation model. • We conduct extensive experiments to evaluate the proposed model and algorithm. [ABSTRACT FROM AUTHOR]

A metaheuristic algorithm called Harris hawks optimization (HHO) is gaining its popularity among its clan and useful for optimization. In this algorithm, the prey gets completely exhausted when the escape energy is equal to zero, therefore it fails to explore further. The random operator chosen in the existing method is a wastage of search agents (Harris hawk). To overcome this issue, we propose an adaptive Harris Hawks optimization (AHHO) technique. In this work, the mutation is employed to restrict the escape energy within the range 0 , 2 , except for the mutation interval. Our method adaptively decides the chance of the Harris hawk would do perch along with the other family members or move to a random tall tree with the help of average fitness. The proposed AHHO algorithm is benchmarked with 23 classical test functions and 30 modern test function from CEC 2014 test suite consisting of unimodal, multimodal, hybrid and composite functions. The qualitative and quantitative analysis, which include metrics such as statistical results, convergence curves, p -value from Wilcoxon rank-sum test and Friedman mean rank. It reveals that AHHO provides good results when compared with other well-known nature-inspired algorithms. It can be used for multilevel thresholding which is an optimization problem. Recently, 2D histogram based multilevel image thresholding techniques are becoming more popular for different image processing applications. The local averaging scheme used for the construction of a 2D histogram in existing methods fails to preserve the edge information. The choice of the diagonal pixels only results in the loss of information making the earlier multi-level thresholding methods inefficient to retain the spatial correlation information. Although the computation of 2D histogram based on grey gradient information is a better way to threshold an image, it faces problems due to the presence of the edge magnitude peaks. These problems are solved by investigating an improved 2D grey gradient (I2DGG) method, a new technique is suggested in this paper to suppress high edge magnitudes. The I2DGG is a maximization problem, which requires an exhaustive search process. Therefore, AHHO is used to obtain the optimal threshold values. The result of our proposed AHHO based multilevel thresholding using the I2DGG method is obtained using all the 500 images from the Berkeley Segmentation Data set (BSDS 500). When we compare the proposed method I2DGG with 2D Tsallis entropy and 1D Tsallis entropy based multilevel thresholding, the I2DGG outperforms other methods. The experimental results are also compared with the state-of-art optimization-based multilevel thresholding methods, which shows our proposed method is beneficial to the segmentation field of image processing. • An adaptive Harris hawks optimization technique is proposed. • A novel I2DGG method for multi-level image thresholding is fostered. • It uses newly introduced normalized grey gradient based 2D histogram. • It proposes a new objective function for optimization. • Experimental results on BSDS 500 are presented. [ABSTRACT FROM AUTHOR]

Computational intelligence in finance has been a very popular topic for both academia and financial industry in the last few decades. Numerous studies have been published resulting in various models. Meanwhile, within the Machine Learning (ML) field, Deep Learning (DL) started getting a lot of attention recently, mostly due to its outperformance over the classical models. Lots of different implementations of DL exist today, and the broad interest is continuing. Finance is one particular area where DL models started getting traction, however, the playfield is wide open, a lot of research opportunities still exist. In this paper, we tried to provide a state-of-the-art snapshot of the developed DL models for financial applications. We not only categorized the works according to their intended subfield in finance but also analyzed them based on their DL models. In addition, we also aimed at identifying possible future implementations and highlighted the pathway for the ongoing research within the field. • All searchable articles of deep learning (DL) for financial applications are reviewed. • DL for finance studies based on their application areas were clustered. • DL models according to their performances in different implementation areas were compared. • To best of our knowledge, this is the first comprehensive DL survey for financial applications. • Current status of DL in finance was provided, also the future opportunities were highlighted. [ABSTRACT FROM AUTHOR]

Computational intelligence methods have gained importance in several real-world domains such as process optimization, system identification, data mining, or statistical quality control. Tools are missing, which determine the performance of computational intelligence methods in these application domains in an objective manner. Statistics provide methods for comparing algorithms on certain data sets. In the past, several test suites were presented and considered as state of the art. However, there are several drawbacks of these test suites, namely: (i) problem instances are somehow artificial and have no direct link to real-world settings; (ii) since there is a fixed number of test instances, algorithms can be fitted or tuned to this specific and very limited set of test functions; (iii) statistical tools for comparisons of several algorithms on several test problem instances are relatively complex and not easily to analyze. We propose a methodology to overcome these difficulties. It is based on standard ideas from statistics: analysis of variance and its extension to mixed models. This paper combines essential ideas from two approaches: problem generation and statistical analysis of computer experiments. • Random selection of test problems plus mixed-model ANOVA enables generalization. • Overcomes over-fitting on fixed artificial test problems. • Suitable for both single and multi-algorithm performance on whole problem classes. • Pairwise algorithm performance comparison with confidence intervals. [ABSTRACT FROM AUTHOR]

Segmenting small-scale residential solar panels (RSPs) based on satellite images is an emerging data science problem in the renewable energy field. In this paper, we develop a cross learning driven U-Net (CrossNets) method and its extension, adaptive CrossNets, to automatically segment RSPs in satellite images. Proposed methods employ a group of generic U-Nets as a community and target to enhance the RSP segmentation performance. First, parameters of each generic U-Net in the community of CrossNets are initialized individually via the initialization with transfer learning and the classical initialization methods. Next, a novel training mechanism, cross learning, is developed to serve as a constraint for better optimizing CrossNets. Based on cross learning, each generic U-Net in the community first individually updates parameters at every epoch and next learns parameters from the best individual at specific epochs. Cross learning relieves the reliance of generic U-Nets on a careful initialization and better optimizes U-Nets. In testing, the result of the best performed generic U-Net in the community is selected as the final segmentation result of CrossNets. Adaptive CrossNets, a variant of CrossNets, is developed by applying an additional threshold to reduce the possibility of over-learning caused by cross learning. Satellite images collected from one city in U.S. are utilized to validate the performance of proposed methods. These images cover a large area of 135 km2 with 2794 RSPs. Compared with two generic U-Nets based benchmarks, our method can enhance the overall segmentation IoU by around 34% and 1.5%. Moreover, the segmentation robustness is improved from 1.191e−2 and 1.286e−4 to 2.481e−5. In addition, two new image datasets collected from other two cities in U.S. are applied to further examine the applicability of proposed methods. • An automatic segmentation of residential solar panels from satellite images is studied. • A cross learning driven U-net method and its adaptive version are developed. • Effectiveness of developed cross learning U-nets on segmenting solar panels in satellite images is evaluated. • Three sets of satellite images are utilized in computational experiments. • The developed method offers a new option for surveying the solar energy utilization in residential regions. [ABSTRACT FROM AUTHOR]