This work proposes a hierarchical approach to reduce the training time of task-based routines by reusing previously obtained autotuning information. This approach has been integrated into a working prototype of Chameleon, a dense linear algebra software whose tile-based routines are executed on the available computational resources by means of a runtime system. The results show that this approach provides a high degree of scalability to the entire self-optimization process, achieving a reduction in training time of up to 80

Over the last two decades, semantic ontologies have been developed to represent manufacturing data across various domains. These ontologies constitute the knowledge base of manufacturing management systems, which primarily focus on optimizing the manufacturing process and improving its resilience. The ontologies developed in the Industry 4.0 domain are heterogeneous, hindering the interoperability of machines, devices, and applications composing manufacturing systems. Consequently, a demand arises for an ontology that provides common vocabularies to represent the data domains inherent to Industry 4.0. A global Industry 4.0 ontology must be easily reusable in different application contexts. This paper presents I40GO: a global ontology tailored to the Industry 4.0 domain. I40GO structures in layers and modules the knowledge represented in the Industry 4.0 most relevant ontologies. The MODDALS methodology is followed to classify knowledge into different layers. This methodology classifies ontology knowledge into common, variant, and application-specific layers following a similar approach to that of Software Product Lines (SPL). I40GO assists ontology engineers in developing domain-specific ontologies for manufacturing systems and enhances interoperability among applications. This work provides an overview of I40GO, emphasizing its development methodology and its modular and layered structure. Furthermore, it demonstrates the reuse of the I40GO ontology within an Industry 4.0 use case-an architecture for context-aware workflow management.

This article develops an architecture for the implementation of Artificial Intelligence in the manufacturing value chain based on standard technologies and data spaces. The standards considered are IEC 63278 “Asset Administration Shell (AAS) for industrial applications” and DIN SPEC 27070:2020 – “Requirements and reference architecture of a security gateway for the exchange of industry data and services“ by IDSA. The architecture provides a data space that allows MONDRAGON industrial cooperatives to use data for the execution of advanced data analytics, Artificial Intelligence (AI) algorithms and interoperability between assets and IoT-platforms. The development of knowledge in this field allows, on the one hand, to optimise the consolidation of data as a strategic factor and, on the other hand, to increase collaboration between manufacturing companies, suppliers and technology providers. The article also explores specific Artificial Intelligence technologies with a wide application in industrial environments. In particular, the study has focused on research into Low/No Code, Explainability (XAI) tools and incremental learning algorithms. The contributions of this paper are summarised in 1) creating an IDS-AAS based architecture and data space that allows the exploitation of AI use cases, either by directly downloading models or by using AI as a service, 2) identifying useful AI tools for industry such as AutoML, No/Low code, XAI or incremental learning, 3) implementing a use case where different AI use alternatives are implemented.

Manufacturing processes of the future will rely on standards for asset interoperability and service orchestration. The Asset Administration Shell (AAS) facilitates information exchange among Industry 4.0 assets, while standardized Business Processes enable workflow execution in manufacturing systems. Combining these technologies provides agility and scalability to manufacturing systems by incorporating asset services within business processes. Service orchestration involves coordinating multiple services, which must be dynamic during runtime to manage unforeseen situations that may arise during the manufacturing process. Context information plays a crucial role in identifying such scenarios and selecting the most suitable devices/services in response, and the Semantic Web accurately represents this information. This paper proposes a context-aware approach for service orchestration using industrial asset services. Our contributions include (1) a component for Context-Aware Service Re-Selection. (2) a domain-specific ontology (DeviceServiceOnt) for Semantic Web-based context representation. And, (3) validation of our proposal in a manufacturing setting where robots are responsible for dispatching and distributing materials within a warehouse. Opportunities for future work are also highlighted, with a primary focus on enhancing workflow dynamicity with context-aware capabilities.

In this work, an automatic optimisation approach for parallel routines on multi-GPU systems is presented. Several inter-GPU communication libraries (such as CUDA-Aware MPI or NCCL) are used with a set of routines to perform the numerical operations among the GPUs located on the compute nodes. The main objective is the selection of the most appropriate communication library, the number of GPUs to be used and the workload to be distributed among them in order to reduce the cost of data movements, which represent a large percentage of the total execution time. To this end, a hierarchical modelling of the execution time of each routine to be optimised is proposed, combining experimental and theoretical approaches. The results show that near-optimal decisions are taken in all the scenarios analysed.

This work presents several self-optimization strategies to improve the performance of task-based linear algebra software on heterogeneous systems. The study focuses on Chameleon, a task-based dense linear algebra software whose routines are computed using a tile-based algorithmic scheme and executed in the available computing resources of the system using a scheduler which dynamically handles data dependencies among the basic computational kernels of each linear algebra routine. The proposed strategies are applied to select the best values for the parameters that affect the performance of the routines, such as the tile size or the scheduling policy, among others. Also, parallel optimized implementations provided by existing linear algebra libraries, such as Intel MKL (on multicore CPU) or cuBLAS (on GPU) are used to execute each of the computational kernels of the routines. Results obtained on a heterogeneous system composed of several multicore and multiGPU are satisfactory, with performances close to the experimental optimum.

PARCSIM is a parallel software simulator that allows a user to capture, through a graphical interface, matrix algorithm schemes that solve scientific problems. With this tool, the user can analyse the execution times that would be obtained by using different spatio-temporal mapping of computational tasks on available computational units, parallelism parameters and computational libraries. Furthermore, for complex problem models, the self-optimization engine incorporated in this tool analyses the huge tree of possible calculations grouping and mapping strategies in search of the choice that makes the best use of the available hardware resources. This tool also offers polyalgorithmic resolution by making automatically the best decision between different software approaches to solve a given problem on the hardware system available. This work shows the usefulness of this simulator to efficiently solve hierarchical problems constructed from previously modelled subproblems. This task is performed by reusing, in a scalable way, the optimization information of these subproblems to establish the best execution configuration for the composite problem.

This paper presents an industrial scenario that simulates a Manufacturing as a Service system for the execution of remote production orders built upon the implementation of emerging Asset Administration Shell (AAS) capabilities and International Data Space connectors. Static and dynamic information from industrial assets (presses and laser cutting machines) are modelled with new AAS submodels and the result is stored in an AAS manager/registration system. A manufacturing orchestrator discovers assets through the registry and completes production orders. The AAS registry allows the selection of assets with capabilities to perform tasks and also shares the AAS catalogue available in the system. The catalogue is shared with external parties through Data Space Connectors. Third party companies can launch manufacturing orders remotely using the same connectors. The paper validates the implementation of AAS components and IDS connectors in a manufacturing context where remote production orders can be securely activated.

Microservice Architectures have increasingly become popular in Industry 4.0 as they allow heterogeneous systems to interact, reduce the complexity in the management of individual components, and support distributed deployments. The integration of those distributed services into orchestrated production processes is performed by workflow managers. Next generation workflow managers must overcome a number of challenges when operating in microservice architectures and IoT environments. To overcome these challenges (heterogeneity, high dynamism, edge deployment or scalability), we propose a work-flow manager alternative built in Node-RED. Node-RED provides instruments for the development of IoT systems and leverages the edge computing paradigm. This solution is deployable in embedded systems, is able to load and execute business processes by means of BPMN recipes and enables the integration of other frameworks and architectures.

In recent years, the use of technology is gaining weight in higher education. Today’s students are digital natives and e-Learning is common for them. Furthermore, they find traditional teaching methods tedious. In order to improve their motivation, different strategies are used, for instance, Flipped Classroom instructional model or gamification. One of the fundamental steps of the Flipped Classroom approach is the application of quizzes to assess the knowledge acquired by the students in the theoretical activity that they develop at home. Tools such as Google forms, Moodle or Kahoot can be used for these quizzes. Based on our experience of introducing the Flipped Classroom approach in the computer science degree in 11 subjects, we show the advantages and disadvantages of each of analyzed tools. Class quizzes results show that the theoretical subjects obtained a higher success rate and also in the subjects in which Kahoot was used, the results were higher. In addition, students rated better the Flipped Classroom surveys in which Kahoot was used. Finally, the application of Flipped Classroom approach improved the students’ assessment of those subjects when compared to the previous course.

In the last years, the huge amount of data available in many disciplines makes the mathematical modeling, and, more concretely, econometric models, a very important technique to explain those data. One of the most used of those econometric techniques is the Vector Autoregression Models (VAR) which are multi-equation models that linearly describe the interactions and behavior of a group of variables by using their past. Traditionally, Ordinary Least Squares and Maximum likelihood estimators have been used in the estimation of VAR models. These techniques are consistent and asymptotically efficient under ideal conditions of the data and the identification problem. Otherwise, these techniques would yield inconsistent parameter estimations. This paper considers the estimation of a VAR model by minimizing the difference between the dependent variables in a certain time, and the expression of their own past and the exogenous variables of the model (in this case denoted as VARX model). The solution of this optimization problem is approached through hybrid metaheuristics. The high computational cost due to the huge amount of data makes it necessary to exploit High-Performance Computing for the acceleration of methods to obtain the models. The parameterized, parallel implementation of the metaheuristics and the matrix formulation ease the simultaneous exploitation of parallelism for groups of hybrid metaheuristics. Multilevel and heterogeneous parallelism are exploited in multicore CPU plus multiGPU nodes, with the optimum combination of the different parallelism parameters depending on the particular metaheuristic and the problem it is applied to.

Parallel numerical libraries for modern architectures can be finely optimized. However, this optimization process may require to carefully set up many parameters, which can be a pretty cumbersome task when done manually. For instance, modern dense linear algebra libraries often split the matrix in submatrices whose size may significantly impact performance. Similarly, the scheduling policy may be of importance, especially when dealing with heterogeneous architectures. Possibly, it may also be interesting to let computational units idle. In this work, we discuss how to automatically decide the selection of these parameters. The idea is that, during a preliminary phase at install time, well chosen tests are performed, so that, once the library has been installed, these parameters get automatically (and instantly) decided while achieving an overall performance close to optimum. We illustrate our discussion with the Chameleon dense linear algebra library running on top of the StarPU runtime system. We use the SimGrid simulator to predict the performance of time consuming executions (large matrices) and hence reduce the time spent in the autotuning phase.

Global ontologies include common vocabularies to provide interoperability among different applications. These ontologies require a balance of reusability-usability to minimise the ontology reuse effort in different applications. To achieve such a balance, reusable and usable ontology design methodologies provide guidelines to design and develop layered ontology networks. Layered ontology networks classify into different abstraction layers the domain knowledge relevant to many applications (common domain knowledge) and the domain knowledge relevant only to certain application types (variant domain knowledge). This knowledge classification is performed from scratch by domain experts and ontology engineers. This process is a heavy workload, making it difficult to design the layered structures of reusable and usable global ontologies. Considering how common and variant software features are classified when designing Software Product Lines (SPLs), we argue that SPL engineering techniques can facilitate the domain knowledge classification taking as reference existing ontologies. This paper presents a methodology that provides guidelines to design the layered structure of reusable and usable ontology networks called MODDALS. In contrast to previous methods, MODDALS applies SPL engineering techniques to systematically (1) identify the ontology common and variant domain knowledge and (2) classify it into different abstraction layers taking as reference existing ontologies. This approach complements domain experts' and ontology engineers' expertise, preventing them from classifying the domain knowledge from scratch facilitating the design of the layered ontology structure. MODDALS methodology is evaluated in the design of the layered structure of a reusable and usable global ontology for the energy domain. The results show that MODDALS enables to classify the domain knowledge taking as reference existing ontologies.

The new paradigm of the Industry 4.0 centers on the digitalization of assets to realize a new industrial revolution. Standardization and interoperability are key for the successful implementation of this digitalization strategy. Among the different standardization and interoperability initiatives, Asset Administration Shell (AAS) proposes a standardized electronic representation of industrial assets enabling Digital Twins and interoperability between automated industrial systems and Cyber Physical System (CPS). In this context, Mondragon Corporation has launched several initiatives to boost the digitalization of its industries. Although implementation of the AAS in real industrial scenarios is not widespread, Mondragon Corporation has identified this initiative as a key enabler for manufacturing companies within its group. This paper presents a case study on the application of the AAS in an industrial context. The AAS initiative is implemented through integrating a Machine Tooling ecosystem with a robotic arm. This implementation facilitates the discovery and integration of grinding machines with other components or machines in a production plant, validating the AAS in a manufacturing scenario.

A hierarchical approach for autotuning linear algebra routines on heterogeneous platforms is presented. Hierarchy helps to alleviate the difficulties of tuning parallel routines for high-performance computing systems. This paper analyzes the application of the hierarchical approach at both the hardware and software levels, using the basic matrix multiplication and the Strassen multiplication as proof of concept on multicore+coprocessor nodes. In this way, the hierarchical approach allows partial delegation of the efficient exploitation of the computing units in the node to the underlying direct autotuned matrix multiplication used in the base case.

A roadmap for autotuning task-based numerical libraries is presented. Carefully chosen experiments are carried out when the numerical library is being installed to assess its performance. Real and simulated executions are considered to optimize the routine. The discussion is illustrated with a task-based tile Cholesky factorization, and the aim is to find the optimum tile size for any problem size, using the Chameleon numerical linear algebra package on top of the StarPU runtime system and also with the SimGrid simulator. The study shows that combining a smart exploration strategy of the search space with both real and simulated executions results in a fast, reliable autotuning process.

The heterogeneity of energy ontologies hinders the interoperability between ontology-based energy management applications to perform a large-scale energy management. Thus, there is the need for a global ontology that provides common vocabularies to represent the energy subdomains. A global energy ontology must provide a balance of reusability–usability to moderate the effort required to reuse it in different applications. This paper presents DABGEO: a reusable and usable global ontology for the energy domain that provides a common representation of energy domains represented by existing energy ontologies. DABGEO can be reused by ontology engineers to develop ontologies for specific energy management applications. In contrast to previous global energy ontologies, it follows a layered structure to provide a balance of reusability–usability. In this work, we provide an overview of the structure of DABGEO and we explain how to reuse it in a particular application case. In addition, the paper includes an evaluation of DABGEO to demonstrate that it provides a balance of reusability–usability.

Left ventricular non-compaction is characterized by the presence of multiple trabecules in the left ventricle myocardium, associated with multiple inter-trabecular recesses communicated with the ventricular cavity. The medical community needs an objective quantification of non-compacted cardiomyopathy, characterized by a trabeculated mass in the left ventricle myocardium. A software tool for the automatic quantification of the exact hyper-trabeculation degree in the left ventricle myocardium for a population of hypertrophic cardiomyopathy (QLVTHC) patients is developed and tested. End-diastolic cardiac magnetic resonance images of the patients are the input of the software, while the volumes of the compacted zones and the trabeculated zones are necessary to produce the percentage quantification of the trabecular zone with respect to the compacted zone. Significant improvements are obtained with respect to the manual process, by saving valuable diagnosis time. The development of a self-optimized software tool (SOST) based on the outputs of 50 patients with hypertrophic cardiomyopathy automatically produces the volumes of the compacted zones and the trabeculated zones, as a percentage quantification. Now, the SOST is tested with a different population of patients, with different characteristics. Besides, a parallelization for the detection of the external layer of the compacted zone allows the real-time analysis per slice in a patient, obtaining important speedups with regard to the QLVTHC proposed and the manual process used traditionally by cardiologists.

Current cities address efficiency challenges for optimizing the use of limited resources. City sustainability and resilience must also be improved through new learning and cognitive technologies that change citizen behavioural patterns and react to disruptive changes. These technologies will allow the evolution of current cities towards the so called "Cognitive Cities". This chapter highlights the importance of Semantic Web and semantic ontologies as a foundation for learning and cognitive systems. Energy is one of the city domains where learning and cognitive systems are needed. This chapter reviews Information and Communication Technologies (ICT)-based energy management solutions developed to improve city energy efficiency, sustainability and resilience. The review focuses on learning and cognitive solutions that improve energy sustainability and resilience through Semantic Web technologies. In addition, these solutions are evaluated from level of acceptance and use of semantics perspectives. The evaluation highlights that the Cognitive City approach is in the early stages in the energy domain and demonstrates the need for a standard energy ontology.

Global ontologies must provide a balance of reusabilityusability to minimize the ontology reuse effort in different applications.To achieve this balance, ontology design methods focus on designing layered ontologies that classify into abstraction layers the common domain knowledge (reused by most applications) and the variant domain knowledge (reused by specific application types).This classification is performed from scratch by domain experts and ontology engineers.Hence, the design of reusable and usable ontologies that represent complex domains takes a lot of effort.Considering how common and variant software features are classified when designing Software Product Lines (SPLs), we argue that SPL engineering techniques can facilitate the domain knowledge classification taking as reference existing ontologies.In this paper, we show the experiences of applying SPL and ontology design techniques in combination to design a reusable and usable global ontology for the energy domain.Domain experts and ontology engineers evaluated the proposed method.The results show that SPL engineering techniques enable a systematic and accurate domain knowledge classification, thus saving ontology design effort.