This work addresses the performance comparison between four clustering techniques with the objective of achieving strong hybrid models in supervised learning tasks. A real dataset from a bio-climatic house named Sotavento placed on experimental wind farm and located in Xermade (Lugo) in Galicia (Spain) has been collected. Authors have chosen the thermal solar generation system in order to study how works applying several cluster methods followed by a regression technique to predict the output temperature of the system. With the objective of defining the quality of each clustering method two possible solutions have been implemented. The first one is based on three unsupervised learning metrics (Silhouette, Calinski-Harabasz and Davies-Bouldin) while the second one, employs the most common error measurements for a regression algorithm such as Multi Layer Perceptron.

This special issue compiles recent applications of Computational Intelligence in the security domain. It is aimed at both researchers and practitioners from academia and industry who are engaged in deploying intelligent solutions for securing both information and systems. Four papers are included in this special issue, covering a wide variety of case studies, ranging from assisted driving to explainable robotics. Additionally, a variety of cutting-edge techniques are proposed, such as blockchain and autoencoders among others. All these works contribute to the scientific progress in this domain, going one step further. In the first contribution by Barreno et al., it is proposed a fuzzy expert system aimed at identifying and classifying conventional two-lane roads based on their geometric characteristics and the expected vehicle performance (passenger cars, trucks, and buses). The road geometric features were measured by sensors in an equipped vehicle, travelling in real-life roads located in the Madrid region (Spain). Fuzzy and neuro-fuzzy systems are applied in order to classify road sections based on their current geometric characteristics. Additionally, a risk identification system is presented. Its main target is to assess whether a vehicle is driving on a two-lane road with an inappropriate speed. In order to do that, an identification model is applied to some variables such as the vehicle type, the road longitudinal gradient, the angle covered by each horizontal curve, and the existence or not of an additional lane of traffic. A fuzzy risk index is proposed to verify relatively unsafe road sections, where more accidents are prone to happen. The next contribution, by Wilusz and Wojtowicz, proposes an architecture for cryptocurrency insurance framework along with a set of corresponding communication protocols and a smart contract template that jointly reduce various risks to which cryptocurrency-based transactions are exposed. Some of the threats are unintentional credential leaks, fraudulent cryptocurrency exchanges and broken private keys among others. Authors analyse such risks and present a solution that can potentially enable new business models based on this technology. In the proposed insurance framework, the holder transfers the risk to the insurer and the insurer applies technological countermeasures to secure the cryptocurrency unit. Funds can be transferred with the bilateral agreement of the cryptocurrency unit holder and the insurer thanks to the use of multi-signatures required by smart contracts. This framework is analysed from the security standpoint, guaranteeing that it can be deployed in several practical scenarios in various domains of applications. Zayas-Gato et al. address the challenging and up-to-date issue of detecting anomalies in industrial systems. In order to automatically detect the anomalies, a hybrid system is proposed. First, data are clustered by the Density-Based Spatial Clustering of Applications With Noise (DBSCAN) algorithm. Then, eight one-class techniques are benchmarked to model the normal operation the target industrial plants, namely: NCBoP, Autoencoders, Gauss, K-Centers, Minimum Spanning Trees, Parzen Density Estimator, Principal Component Analysis, and Support Vector Data Description. Authors' proposal is validated using two real-life domains with data gathered during different operating points of two industrial plants. The first case is controlling the liquid level in a tank while the second one is manufacturing wind generator blades made of carbon fibre material. The validated approach can be used to improve the classification when the target set is scattered in different clusters or groups. As a result, an early detection of anomalous situations is more feasible, increasing the systems optimisation. The last contribution, by Rodríguez-Lera et al., focus on accountability of intelligent robots. This implies that any robot is in charge of logging its activities with verifiable evidence so that all actions are traceable and the events triggering the action are identified. The main contributions of this study are an overview, modelling, and formalization of the three perspectives of accountability for autonomous robots; a GUI tool to graphically illustrate the knowledge supported on Conceptual Graphs; and finally, a proof of concept summarizing the information dumped on each level using a set of innovative debugging tools. The findings of this research support the idea that a verbose logging system drives to serious performance issues on an accountability platform, and the detailed information provided is far from explaining the robot's behaviour. Besides, large volumes of data might not be managed properly in specific situations; also, the access to such information sacrifices the privacy of the human-robot interaction. The target is that the robot could incorporate the mechanisms that help manufacturers, client/user and developers to know the reasons that trigger a certain behaviour. The guest editors wish to thank Prof. Jon G. Hall (Editor-in-Chief of the Wiley-Blackwell Journal Expert Systems: The Journal of Knowledge Engineering) for providing the opportunity to edit this special issue. Additional thanks to Prof. Lucia Rapanotti, for her support in managing the editing process. Finally, guest editors would also like to thank the referees who have thoroughly evaluated the papers and the editorial staff for their support. The authors declare no potential conflict of interest.

Abstract The significant advance of Internet of Things in industrial environments has provided the possibility of monitoring the different variables that come into play in an industrial process. This circumstance allows the supervision of the current state of an industrial plant and the consequent decision making possibilities. Then, the use of anomaly detection techniques are presented as a powerful tool to determine unexpected situations. The present research is based on the implementation of one‐class classifiers to detect anomalies in two industrial systems. The proposal is validated using two real datasets registered during different operating points of two industrial plants. To ensure a better performance, a clustering process is developed prior the classifier implementation. Then, local classifiers are trained over each cluster, leading to successful results when they are tested with both real and artificial anomalies. Validation results present in all cases, AUC values above 90%.

Abstract The devastating consequences of climate change have resulted in the promotion of clean energies, being the wind energy the one with greater potential. This technology has been developed in recent years following different strategic plans, playing special attention to wind generation. In this sense, the use of bicomponent materials in wind generator blades and housings is a widely spread procedure. However, the great complexity of the process followed to obtain this kind of materials hinders the problem of detecting anomalous situations in the plant, due to sensors or actuators malfunctions. This has a direct impact on the features of the final product, with the corresponding influence in the durability and wind generator performance. In this context, the present work proposes the use of a distributed anomaly detection system to identify the source of the wrong operation. With this aim, five different one-class techniques are considered to detect deviations in three plant components located in a bicomponent mixing machine installation: the flow meter, the pressure sensor and the pump speed.

One critical point to improve the economic and technical results of every industrial process lies on the fact of achieving a good optimization, and applying a smart maintenance plan. In this context, the tools development for detecting the appearance of any kind of anomaly represents an important challenge. For this reason, the implementation of classifiers for anomaly detection tasks has been a significant trend in the scientific community. However, since the behavior of the potential anomalies that may occur in a plant is unknown, it is necessary to generate artificial outliers to assess these classifiers. This paper proposes the performance checking of different intelligent one-class techniques to detect anomalies in an industrial plant, used to obtain the main material for wind generator blades production. These classifiers are tested using anomaly data generated, giving successful results.

Abstract This research describes a novel approach for fault detection in industrial processes, by means of unsupervised and projectionist techniques. The proposed method includes a visual tool for the detection of faults, its final aim is to optimize system performance and consequently obtaining increased economic savings, in terms of energy, material, and maintenance. To validate the new proposal, two datasets with different levels of complexity (in terms of quantity and quality of information) have been used to evaluate five well‐known unsupervised intelligent techniques. The obtained results show the effectiveness of the proposed method, especially when the complexity of the dataset is high.

A feedback solution for approximate optimal scheduling of switched systems with autonomous subsystems and continuous-time dynamics is presented. The proposed solution is based on policy iteration algorithm which provides the optimal switching schedule. Algorithms for offline, online, and concurrent implementation of the proposed solution are presented. For online and concurrent training, gradient descent training laws are used and the performance of the training laws is analyzed. The effectiveness of the presented algorithms is verified through numerical simulations.

Current advances in the development of mobile and smart devices have generated a growing demand for natural human-machine interaction and favored the intelligent assistant metaphor, in which a single interface gives access to a wide range of functionalities and services. Conversational systems constitute an important enabling technology in this paradigm. However, they are usually defined to interact in semantic-restricted domains in which users are offered a limited number of options and functionalities. The design of multi-domain systems implies that a single conversational system is able to assist the user in a variety of tasks. In this paper we propose an architecture for the development of multi-domain conversational systems that allows: (1) integrating available multi and single domain speech recognition and understanding modules, (2) combining available system in the different domains implied so that it is not necessary to generate new expensive resources for the multi-domain system, (3) achieving better domain recognition rates to select the appropriate interaction management strategies. We have evaluated our proposal combining three systems in different domains to show that the proposed architecture can satisfactory deal with multi-domain dialogs.

In this research, a novel family of learning rules called Beta Hebbian Learning (BHL) is thoroughly investigated to extract information from high-dimensional datasets by projecting the data onto low-dimensional (typically two dimensional) subspaces, improving the existing exploratory methods by providing a clear representation of data's internal structure. BHL applies a family of learning rules derived from the Probability Density Function (PDF) of the residual based on the beta distribution. This family of rules may be called Hebbian in that all use a simple multiplication of the output of the neural network with some function of the residuals after feedback. The derived learning rules can be linked to an adaptive form of Exploratory Projection Pursuit and with artificial distributions, the networks perform as the theory suggests they should: the use of different learning rules derived from different PDFs allows the identification of "interesting" dimensions (as far from the Gaussian distribution as possible) in high-dimensional datasets. This novel algorithm, BHL, has been tested over seven artificial datasets to study the behavior of BHL parameters, and was later applied successfully over four real datasets, comparing its results, in terms of performance, with other well-known Exploratory and projection models such as Maximum Likelihood Hebbian Learning (MLHL), Locally-Linear Embedding (LLE), Curvilinear Component Analysis (CCA), Isomap and Neural Principal Component Analysis (Neural PCA).

Dealing with distributed data is one of the challenges for clustering, as most clustering techniques require the data to be centralized. One of them, k-means, has been elected as one of the most influential data mining algorithms for being simple, scalable, and easily modifiable to a variety of contexts and application domains. However, exact distributed versions of k-means are still sensitive to the selection of the initial cluster prototypes and require the number of clusters to be specified in advance. Additionally, preserving data privacy among repositories may be a complicating factor. In order to overcome k-means limitations, two different approaches were adopted in this paper: the first obtains a final model identical to the centralized version of the clustering algorithm and the second generates and selects clusters for each distributed data subset and combines them afterwards. It is also described how to apply the algorithms compared while preserving data privacy. The algorithms are compared experimentally from two perspectives: the theoretical one, through asymptotic complexity analyses, and the experimental one, through a comparative evaluation of results obtained from a collection of experiments and statistical tests. The results obtained indicate which algorithm is more suitable for each application scenario.