Search

Electricity demand is shown to steadily increase in the last few years, and it is one of the key aspects of living standards and quantifying welfare effects. However, the irregularity of electricity demand is one of the main problems in this field. Therefore, it is important to accurately anticipate future expenditures in order to optimize energy generation and to avoid unexpected wastes. As a result, we developed Machine Learning models to predict electricity demand. In particular, our study has been performed using data of the Spanish Electricity Network from 2007 to 2019. To this end, we propose the implementation of a set of Machine Learning techniques using various frameworks. In particular, we implemented six different prediction models: Linear Regression, Regression Trees, Gradient Boosting Regression, Random Forests, Multi-layer Perceptron, and three types of recurrent neural networks. Our experimentation shows promising results in all cases, since our models provides better prediction than the one estimated by the Spanish Electricity Network with an improvement of 12% in the worst case and up to 37% for the best predictor, which turned out to be the Gated Recurrent Unit neural network.

The increase of energy consumption and their direct effects on pollution and global warming have motivated governments to develop new strategies to promote a better usage of energy. One of the most important aspects related to energy efficiency is the need for a suitable model of energy consumption that can be used to make predictions or to aid experts in high level decision making processes. Symbolic regression techniques can be used to discover an energy consumption model that fits these purposes. Traditionally, the problem of symbolic regression has been solved by using genetic programming approaches to find the algebraic expression that best fits the regression problem data, where each expression is encoded as a tree structure. In previous works, we found that a different approach using Straight Line Programs as a representation technique could provide promising results for symbolic regression, although the size of the resulting algebraic expression might be increased when compared to the traditional approach. This work proposes an Ant Colony Optimization algorithm for Straight Line Programs to solve the problem, and makes a study to compare the approach with traditional genetic programming in a real energy consumption modelling problem.

The interfaces between users and systems are evolving into a more natural communication, including user gestures as part of the interaction, where air-writing is an emerging application for this purpose. The aim of this work is to propose a new air-writing system based on only one array of ultrasonic transceivers. This track will be obtained based on the pairwise distance of the hand marker with each transceiver. After acquiring the track, different deep learning algorithms, such as long short-term memory (LSTM), convolutional neural networks (CNN), convolutional autoencoder (ConvAutoencoder), and convolutional LSTM have been evaluated for character recognition. It has been shown how these algorithms provide high accuracy, where the best result is extracted from the ConvLSTM, with 99.51% accuracy and 71.01 milliseconds of latency. Real data were used in this work to evaluate the proposed system in a real scenario to demonstrate its high performance regarding data acquisition and classification., Project “SEMULIN” (German project number 19A20012D), German Federal Ministry for Economic Affairs and Energy (BMWi), Project P20-00265 funded by the “Junta de Andalucia” of Spain

Energy consumption has increased in recent decades at a rate ranging from 1.5% to 10% per year in the developed world. As a consequence, several efforts have been made to model energy consumption in order to achieve a better use of energy and to minimize environmental impact. Open problems in this area range from energy consumption forecasting to user profile mining, energy source planning, to transportation, among others. To address these problems, it is important to have suitable tools to model energy consumption data series, so that the analysts and CEOs can have knowledge about the underlying properties of the power demand in order to make high-level decisions. In this paper, we focus on the problem of energy consumption modelling, and provide a solution from the perspective of symbolic regression. More specifically, we develop hybrid genetic programming algorithms to find the algebraic expression that best models daily energy consumption in public buildings at the University of Granada as a testbed, and compare the benefits of Straight Line Programs with the classic tree representation used in symbolic regression. Regarding algorithm design, the outcomes of our experimentation suggest that Straight Line Programs outperform other representation models in the symbolic regression problems studied, and also that the hybridation with local search methods can improve the quality of the resulting algebraic expression. On the other hand, with regards to energy consumption modelling, our approach empirically demonstrates that symbolic regression can be a powerful tool to find underlying relationships between multivariate energy consumption data series.

This work was supported in part by the Project A-SWARM through the German Federal Ministry of Economy and Industry (BMWI) by the Maritime Forschungsstrategie 2025 under Project 03SX485D., Data acquisition and treatment are key issues for any Deep Learning (DL) technique, especially in computer vision tasks. A big effort must be done for the creation of labeled datasets, due to the time this task requires and its complexity in cases where different sensors must be used. This is the case of radar imaging applications, where radar data are dif cult to analyze and must be labeled manually. In this paper, a semi-automatic framework to generate labels for range Doppler maps (radar images) is proposed. This technique is based on a sensor fusion approach with radar and camera sensors. The proposed scheme operates in two steps: The rst step is the environment features extraction, in which the radar data is preprocessed and ltered to remove ghost targets and detect clusters, and camera data are used to extract the information of the targets. In the second step, a rule-based system that considers the extracted features fuses the information to generate labels for the radar data. By using the proposed framework, the experimentation performed suggests that the time required to label the data is reduced as well as the possibility of human error during the labeling task. Our results show that the proposed technique can improve the nal model accuracy with regards the traditional labeling method, carried out by human experts., Project A-SWARM through the German Federal Ministry of Economy and Industry (BMWI) by the Maritime Forschungsstrategie 2025 03SX485D

Research partially supported by grant PID2019-110525GB-I00 from Agencia Estatal de Investigacion (AEI) and from Fondo Europeo de Desarrollo Regional (FEDER), and by grant A-FQM-470-UGR18 from Consejeria de Economia y Conocimiento de la Junta de Andalucia and Programa Operativo FEDER 2014-2020., Finding the minimum distance of linear codes is an NP-hard problem. Traditionally, this computation has been addressed by means of the design of algorithms that find, by a clever exhaustive search, a linear combination of some generating matrix rows that provides a codeword with minimum weight. Therefore, as the dimension of the code or the size of the underlying finite field increase, so it does exponentially the run time. In this work, we prove that, given a generating matrix, there exists a column permutation which leads to a reduced row echelon form containing a row whose weight is the code distance. This result enables the use of permutations as representation scheme, in contrast to the usual discrete representation, which makes the search of the optimum polynomial time dependent from the base field. In particular, we have implemented genetic and CHC algorithms using this representation as a proof of concept. Experimental results have been carried out employing codes over fields with two and eight elements, which suggests that evolutionary algorithms with our proposed permutation encoding are competitive with regard to existing methods in the literature. As a by-product, we have found and amended some inaccuracies in the Magma Computational Algebra System concerning the stored distances of some linear codes., Agencia Estatal de Investigacion (AEI) PID2019-110525GB-I00, European Commission, Junta de Andalucia A-FQM-470-UGR18, Programa Operativo FEDER 2014-2020

There have been significant advances regarding target detection in the autonomous vehicle context. To develop more robust systems that can overcome weather hazards as well as sensor problems, the sensor fusion approach is taking the lead in this context. Laser Imaging Detection and Ranging (LiDAR) and camera sensors are two of the most used sensors for this task since they can accurately provide important features such as target´s depth and shape. However, most of the current state-of-the-art target detection algorithms for autonomous cars do not take into consideration the hardware limitations of the vehicle such as the reduced computing power in comparison with Cloud servers as well as the reduced latency. In this work, we propose Edge Computing Tensor Processing Unit (TPU) devices as hardware support due to their computing capabilities for machine learning algorithms as well as their reduced power consumption. We developed an accurate and small target detection model for these devices. Our proposed Multi-Level Sensor Fusion model has been optimized for the network edge, specifically for the Google Coral TPU. As a result, high accuracy results are obtained while reducing the memory consumption as well as the latency of the system using the challenging KITTI dataset.

This work addresses the problem of energy consumption time series forecasting. In our approach, a set of time series containing energy consumption data is used to train a single, parameterised prediction model that can be used to predict future values for all the input time series. As a result, the proposed method is able to learn the common behaviour of all time series in the set (i.e., a fingerprint) and use this knowledge to perform the prediction task, and to explain this common behaviour as an algebraic formula. To that end, we use symbolic regression methods trained with both single- and multi-objective algorithms. Experimental results validate this approach to learn and model shared properties of different time series, which can then be used to obtain a generalised regression model encapsulating the global behaviour of different energy consumption time series., This work was supported by the Spanish Government (research project TIN201564776-C3-1-R). M. Molina-Solana was funded by European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 743623.

Energy efficiency in public buildings has become a major research field, due to the impacts of the energy consumption in terms of pollution and economic aspects. For this reason, governments know that it is necessary to adopt measures in order to minimize the environmental impact and saving energy. Technology advances of the last few years allow us to monitor and control the energy consumption in buildings, and become of great importance to extract hidden knowledge from raw data and give support to the experts in decision-making processes to achieve real energy saving or pollution reduction among others. Prediction techniques are classical tools in machine learning, used in the energy efficiency paradigm to reduce and optimize the energy using. In this work we have used two prediction techniques, symbolic regression and neural networks, with the aim of predict the energy consumption in public buildings at the University of Granada. This paper concludes that symbolic regression is a promising and more interpretable results, whereas neural networks lack of interpretability take more computational time to be trained. In our results, we conclude that there are no significant differences in accuracy considering both techniques in the problems addressed.

This article explores the use of multi-objective evolutionary machine learning techniques to find the minimum number of sensors from a pull of 6 sensors as well as the minimum number of analytical signals belonging to each selected sensor for the design of an optimal colourimetric temperature sensor. The analytical information was obtained with a calibrated neural network that provides the best temperature estimation with respect to the selected colourimetric sensor responses from a previously developed sensor array. The sensor array was developed by embedding the linear spin crossover material [Fe-(NH 2 trz) 3 ](BF 4 ) 2 into polymers with different polarity, offering different thermochromic responses related to different morphologies of the spin crossover particles when embedded in each polymer. The different thermochromic responses are tracked by the green component of the RGB colour space and the a * from CIEL* a * b * obtained with a conventional photographic digital camera. These two colour signals are used as analytical parameters for the subsequent computer processing and model calibration. The use of multi-objective optimization techniques for neural network calibration demonstrated that only 3 signals coming from 3 sensors of the 6 studied are sufficient to provide optimal temperature estimation. The optimized selection was the green channel from polyurethane hydrogel D6 and PVC prepared in THF and a * from PMMA prepared in toluene.

International Conference on Hybrid Artificial Intelligence Systems (2017. Logroño), The estimation of the future streamflows is one of the main research topics in hydrology and a very important task for water resources management. The aim of this work is to use symbolic regression in order to model the hydrological balance. Specifically, we use genetic programming to solve the symbolic regression problem. Nevertheless, in this work we use Straight Line Programs instead of trees to encode algebraic expression. Results shows that this representation for algebraic expressions could improve the results in both accuracy and computational time., Departamento de Ciencias de la Computación e Inteligencia Artificial, Universidad de Granada, España, Departamento de Ingeniería Civil, Universidad Católica de San Antonio, España, Unidad de Granada, Instituto Geológico y Minero de España, España

Describing user activity plays an essential role in ambient intelligence. In this work, we review different methods for human activity recognition, classified as data-driven and knowledge-based techniques. We focus on context ontologies whose ultimate goal is the tracking of human behavior. After studying upper and domain ontologies, both useful for human activity representation and inference, we establish an evaluation criterion to assess the suitability of the different candidate ontologies for this purpose. As a result, any missing features, which are relevant for modeling daily human behaviors, are identified as future challenges.

Thermochromic thin films of the spin crossover (SCO) polymer [Fe(NH2trz)3](BF4) are prepared using a variety of organic polymers as hosts. The formation of different polymeric networks is confirmed macroscopically by the colour changes related to an SCO phenomenon induced by thermal variation, and the results are correlated with electron microscopy and energy dispersive X-ray spectroscopy. Large particles of the SCO material are observed in SCO/polymer hybrid systems with hydrophobic polymers, while more dispersed nano-crystals appear in the hydrophilic matrices, leading to the transformation of the particles into fibrous structures. Subsequently, submicrometer-size SCO fibrous nanoparticles undergo colourimetric spin transitions near room temperature while grains with sizes larger than several microns move their transitions to lower temperatures. The difference in properties between the SCO/polymer hybrid materials is not only due to the differences in the size and shape of the SCO crystals in each polymer but also to the nature of the polymer and solvent interactions. The optical changes obtained for each SCO/polymer hybrid material are related to the microscopic origin of the cooperative interactions tracked by using a photographic digital camera. A linear correlation is obtained (colour values versus temperature) when processing all the colourimetric data by artificial neural networks, thus avoiding the uncertainty inherent in the hysteresis loop.

Smart Homes are intelligent spaces that contain resources to collect information about user's activities and ease the assisted living. Abnormal behavior detection has been remarked as one of the most challenging application fields in this research area, due to its possibilities for assisting elders or people with special needs. These systems help to maintain people's independence, enhancing their personal comfort and safety and delaying the process of moving to a nursing home. In this paper, we describe a new approach for the behavior recognition problem based on Learning Automata and fuzzy temporal windows. Our proposal learns the normal behaviors, and uses that knowledge to recognise normal and abnormal human activities in real time. In addition, our proposal is able to adapt online to environmental variations, changes in human habits, and temporal information, defined as an interval of time when the behavior should be performed.

This paper addresses the problem of energy consumption prediction using neural networks over a set of public buildings. Since energy consumption in the public sector comprises a substantial share of overall consumption, the prediction of such consumption represents a decisive issue in the achievement of energy savings. In our experiments, we use the data provided by an energy consumption monitoring system in a compound of faculties and research centers at the University of Granada, and provide a methodology to predict future energy consumption using nonlinear autoregressive (NAR) and the nonlinear autoregressive neural network with exogenous inputs (NARX), respectively. Results reveal that NAR and NARX neural networks are both suitable for performing energy consumption prediction, but also that exogenous data may help to improve the accuracy of predictions., Department of Computer Science and Artificial Intelligence of the University of Granada TIC111 TIN201564776-C3-1-R

The design of optical sensor arrays encompasses tasks such as the acquisition of sensor optical parameters, as for example color features, and the calibration of a multivariate method to model the array global behavior. Different techniques have been used to model the sensor array responses, such as multivariate regression or neural networks, although they show certain limitations. The former methods require either high amount of computer memory or speed so therefore they are not suitable for implementation in portable electronic devices with low resources, while the later is a black box whose operation cannot be easily modeled and explained for industrial development validation. This work addresses these problems and proposes an expert system to overcome the previous drawbacks. The approach makes an accurate pH prediction, and it comprises a balance between memory and microprocessor speed for its integration within embedded systems with low memory and chip resources. In addition, it is also able to provide a high expressive explanation of why such prediction was made, so that the industrial validation is easier than using other proposals such as neural networks.

A disposable optical tongue has been developed with a two-sensor array prepared on a transparent support. The array consists of polymeric membranes working by ionophore–chromoionophore chemistry with non-specific behaviour controlled by the crown ether-type ionophore present. The system was used for the simultaneous determination of Na(I) and K(I) ions in natural waters. The imaging of the element – after reaction for 3 min with the solution – by a conventional scanner makes it possible to calculate the hue coordinate H of the HSV colour space used as the analytical parameter. The signals were processed using an optimized multilayer Artificial Neural Network (ANN). The individual sensing membranes show good precision (0.3% RSD) and lifetime (around 45 days in darkness). The tongue works over a wide dynamic range (1.0 × 10 −4 to 0.1 M both in Na(I) and K(I)) and when the procedure was applied, the Mean Square Error obtained by the ANN approach was 0.0064 for the prediction of K(I) and 0.0451 for Na(I). The procedure was used to analyze Na(I) and K(I) in different types of natural waters (tap and mineral), validating the results against a reference procedure.

Internet social networks have arisen in the last years as powerful tools where people exchange knowledge and multimedia content. They help to share interests between groups of people with common features. Undoubtedly, there is an inherent social network in any e-learning system, where the main actors are teachers, learners and learning resources. Most e-learning software are mainly focused in content dissemination and group work, but the possibilities that Internet LMSs could offer go further. Recently, there has been research work focused on Web Communities for learning and their formulation as Social Networks. Thus, social network analysis may be applied to infer group structures and to make intelligent recommendation systems and data mining. This paper proposes a method for the formulation and interpretation of learning management platforms as social networks. In order to achieve a major generalization, we develop an ontology to integrate the information from different Learning Management Systems. After that, a personalized social network is extracted from the ontology. This change in the point of view of a LMS could be a challenge to make further studies about learners, teachers and learning resources to obtain a better understanding of their social structure, and therefore to make or improve decisions about the learning process.

We provide a set of projects to put in practice artificial intelligence techniques using LEGO Mindstorms in an undergraduate computer degree, covering reactive and deliberative agents, rule-based systems, graph search algorithms, and planning methods. The projects have been applied for teaching in a third-year undergraduate subject of a computer engineering degree at the University of Granada (Spain). After the contextualization and development of the projects, we discuss the results, advantages, and drawbacks of our experience. © 2011 Wiley Periodicals, Inc. Comput Appl Eng Educ 22: 153–166, 2014

A neural network multivariate calibration is used to predict the pH of a solution in the full-range (0–14) from hue (H) values coming from imaging an optical pH sensor array based on 11 sensing elements with immobilized pH indicators. Different colorimetric acid-base indicators were tested for membrane preparation fulfilling the following conditions: 1) no leaching; 2) change in tonal coordinate by reaction and 3) covering the full pH range with overlapping between their pH responses. The sensor array was imaged after equilibration with a solution using a scanner working in transmission mode. Using software developed by us, the H coordinate of the colour space HSV was calculated from the RGB coordinates of each element. The neural network was trained with the calibration data set using the Levenberg–Marquardt training method. The network structure has 11 input neurons (each one matching the hue of a single element in the sensor array), 1 output (the pH approximation value) and 1 hidden layer with 10 hidden neurons. The network provides an MSE = 0.0098 in the training data, MSE = 0.0183 in the validation data and MSE = 0.0426 in the test data coming from a set of real water samples. The resulting correlation coefficient R obtained in the Pearson correlation test is R = 0.999.

Internet Learning Management Systems (LMSs) are powerful tools that help us in our daily teaching and learning activities. Most users and software are mainly focused in content dissemination and group works, but the possibilities that Internet LMSs could offer go further. Some recent approaches use semantic web to improve the capabilities and user experiences in e-learning by mean of artificial intelligence and knowledge management techniques. In this work, we develop a procedure to achieve the integration of different e-learning systems, and to give semantics to entities and relations in the database of LMSs by mean of ontologies. This integration could ease the dissemination of learning resources and knowledge from the databases of the Learning Management Systems. Moreover, the semantic interpretation of database schemes would allow to find precise information quickly.