Search

This paper presents, for the first time, a distance metric learning algorithm for monotonic classification. Monotonic datasets arise in many real-world applications, where there exist order relations in the input and output variables, and the outputs corresponding to ordered pairs of inputs are also expected to be ordered. Monotonic classification can be addressed through several distance-based classifiers that are able to respect the monotonicity constraints of the data. The performance of distance-based classifiers can be improved with the use of distance metric learning algorithms, which are able to find the distances that best represent the similarities among each pair of data samples. However, learning a distance for monotonic data has an additional drawback: the learned distance may negatively impact the monotonic constraints of the data. In our work, we propose a new model for learning distances that does not corrupt these constraints. This methodology will also be useful in identifying and discarding non-monotonic pairs of samples that may be present in the data due to noise. The experimental analysis conducted, supported by a Bayesian statistical testing, demonstrates that the distances obtained by the proposed method can enhance the performance of several distance-based classifiers in monotonic problems. [ABSTRACT FROM AUTHOR]

Land-Use and Land-Cover (LULC) mapping is relevant for many applications, from Earth system and climate modelling to territorial and urban planning. Global LULC products are continuously developing as remote sensing data and methods grow. However, there still exists low consistency among LULC products due to low accuracy in some regions and LULC types. Here, we introduce Sentinel2GlobalLULC, a Sentinel-2 RGB image dataset, built from the spatial-temporal consensus of up to 15 global LULC maps available in Google Earth Engine. Sentinel2GlobalLULC v2.1 contains 194877 single-class RGB image tiles organized into 29 LULC classes. Each image is a 224 × 224 pixels tile at 10 × 10 m resolution built as a cloud-free composite from Sentinel-2 images acquired between June 2015 and October 2020. Metadata includes a unique LULC annotation per image, together with level of consensus, reverse geo-referencing, global human modification index, and number of dates used in the composite. Sentinel2GlobalLULC is designed for training deep learning models aiming to build precise and robust global or regional LULC maps. Measurement(s) land cover • land use Technology Type(s) satellite imaging of a planet [ABSTRACT FROM AUTHOR]

Data stream mining extracts information from large quantities of data flowing fast and continuously (data streams). They are usually affected by changes in the data distribution, giving rise to a phenomenon referred to as concept drift. Thus, learning models must detect and adapt to such changes, so as to exhibit a good predictive performance after a drift has occurred. In this regard, the development of effective drift detection algorithms becomes a key factor in data stream mining. In this work we propose CURIE , a drift detector relying on cellular automata. Specifically, in CURIE the distribution of the data stream is represented in the grid of a cellular automata, whose neighborhood rule can then be utilized to detect possible distribution changes over the stream. Computer simulations are presented and discussed to show that CURIE , when hybridized with other base learners, renders a competitive behavior in terms of detection metrics and classification accuracy. CURIE is compared with well-established drift detectors over synthetic datasets with varying drift characteristics. [ABSTRACT FROM AUTHOR]

The17 Sustainable Development Goals (SDGs) established by the United Nations Agenda 2030 constitute a global blueprint agenda and instrument for peace and prosperity worldwide. Artificial intelligence and other digital technologies that have emerged in the last years, are being currently applied in virtually every area of society, economy and the environment. Hence, it is unsurprising that their current role in the pursuance or hampering of the SDGs has become critical. This study aims at providing a snapshot and comprehensive view of the progress made and prospects in the relationship between artificial intelligence technologies and the SDGs. A comprehensive review of existing literature has been firstly conducted, after which a series SWOT (Strengths, Weaknesses, Opportunities and Threats) analyses have been undertaken to identify the strengths, weaknesses, opportunities and threats inherent to artificial intelligence-driven technologies as facilitators or barriers to each of the SDGs. Based on the results of these analyses, a subsequent broader analysis is provided, from a position vantage, to (i) identify the efforts made in applying AI technologies in SDGs, (ii) pinpoint opportunities for further progress along the current decade, and (iii) distill ongoing challenges and target areas for important advances. The analysis is organized into six categories or perspectives of human needs: life, economic and technological development, social development, equality, resources and natural environment. Finally, a closing discussion is provided about the prospects, key guidelines and lessons learnt that should be adopted for guaranteeing a positive shift of artificial intelligence developments and applications towards fully supporting the SDGs attainment by 2030. [ABSTRACT FROM AUTHOR]

Foreign exchange (FOREX) market is a decentralized global marketplace in which different participants, such as international banks, companies or investors, can buy, sell, exchange and speculate on currencies. This market is considered to be the largest financial market in the world in terms of trading volume. Indeed, the just-in-time price prediction for a currency pair exchange rate (e.g., EUR/USD) provides valuable information for companies and investors as they can take different actions to improve their business. The trading volume in the FOREX market is huge, disperses, in continuous operations (24 h except weekends), and the context significantly affects the exchange rates. This paper introduces a context-aware algorithm to model the behavior of the FOREX Market, called parallel elastic network model (PENM). This algorithm is inspired by natural procedures like the behavior of macromolecules in dissolution. The main results of this work include the possibility to represent the market evolution of up to 21 currency pair, being all connected, thus emulating the real-world FOREX market behavior. Moreover, because the computational needs required are highly costly as the number of currency pairs increases, a hybrid parallelization using several shared memory and message passing algorithms studied on distributed cluster is evaluated to achieve a high-throughput algorithm that answers the real-time constraints of the FOREX market. The PENM is also compared with a vector autoregressive (VAR) model using both a classical statistical measure and a profitability measure. Specifically, the results indicate that PENM outperforms VAR models in terms of quality, achieving up to 930 × speed-up factor compared to traditional R codes using in this field. [ABSTRACT FROM AUTHOR]

Big Data applications are emerging during the last years, and researchers from many disciplines are aware of the high advantages related to the knowledge extraction from this type of problem. However, traditional learning approaches cannot be directly applied due to scalability issues. To overcome this issue, the MapReduce framework has arisen as a 'de facto' solution. Basically, it carries out a 'divide-and-conquer' distributed procedure in a fault-tolerant way to adapt for commodity hardware. Being still a recent discipline, few research has been conducted on imbalanced classification for Big Data. The reasons behind this are mainly the difficulties in adapting standard techniques to the MapReduce programming style. Additionally, inner problems of imbalanced data, namely lack of data and small disjuncts, are accentuated during the data partitioning to fit the MapReduce programming style. This paper is designed under three main pillars. First, to present the first outcomes for imbalanced classification in Big Data problems, introducing the current research state of this area. Second, to analyze the behavior of standard pre-processing techniques in this particular framework. Finally, taking into account the experimental results obtained throughout this work, we will carry out a discussion on the challenges and future directions for the topic. [ABSTRACT FROM AUTHOR]

Estimates of lava volume, and thus effusion rate, are critical for assessing volcanic hazard and are a priority for volcano observatories with responsibility for monitoring. The choice of specific methods used to approximate lava volume depends on both volcanological and practical considerations; in particular, whether field measurements are possible and how often they can be repeated. Volcán El Reventador (Ecuador) is inaccessible, and field measurements can only be made infrequently at a few locations in its caldera. We present both planimetric field and topographic satellite radar-based measurements of lava flow thicknesses and volumes for activity at El Reventador between 2002 and 2009. Lava volumes estimates range from 75 ± 24 × 10 m (based on field measurements of flow thickness) to 90 ± 37 × 10 m (from satellite radar retrieval of flow thickness), corresponding to time-averaged effusion rates of 9 ± 4 m/s and 7 ± 2 m/s, respectively. Detailed flow mapping from aerial imagery demonstrate that lava effusion rate was at its peak at the start of each eruption phase and decreased over time. Measurements of lava thickness made from a small set of Synthetic Aperture Radar (SAR) interferograms allowed the retrieval of the shape of the compound lava flow field and show that in 2009 it was subsiding by up to 6 cm/year. Satellite radar measurements thus have the potential to be a valuable supplement to ground-based monitoring at El Reventador and other inaccessible volcanoes. [ABSTRACT FROM AUTHOR]

This paper focuses on the use of multi-objective evolutionary algorithms to develop smartly tuned fuzzy logic controllers dedicated to the control of heating, ventilating and air conditioning systems, energy performance, stability and indoor comfort requirements. This problem presents some specific restrictions that make it very particular and complex because of the large time requirements needed to consider multiple criteria (which enlarge the solution search space) and the long computation time models required in each evaluation. In this work, a specific multi-objective evolutionary algorithm is proposed to obtain more compact fuzzy logic controllers as a way of finding the best combination of rules, thus improving the system performance to better solve the HVAC system control problem. This method combines lateral tuning of the linguistic variables with rule selection. To this end, two objectives have been considered, maximizing the performance of the system and minimizing the number of rules obtained. This algorithm is based on the well-known SPEA2 but uses different mechanisms for guiding the search towards the desired Pareto zone. Moreover, the method implements some advanced concepts such as incest prevention, that help to improve the exploration/exploitation trade-off and consequently its convergence ability. The proposed method is compared to the most representative mono-objective steady-state genetic algorithms previously applied to the HVAC system control problem, and to generational and steady-state versions of the most interesting multi-objective evolutionary algorithms (never applied to this problem) showing that the solutions obtained by this new approach dominate those obtained by these methods. The results obtained confirm the effectiveness of our approach compared with the rest of the analyzed methods, obtaining more accurate fuzzy logic controllers with simpler models. [ABSTRACT FROM AUTHOR]

In recent years, there has been a growing interest for the experimental analysis in the field of evolutionary algorithms. It is noticeable due to the existence of numerous papers which analyze and propose different types of problems, such as the basis for experimental comparisons of algorithms, proposals of different methodologies in comparison or proposals of use of different statistical techniques in algorithms’ comparison. In this paper, we focus our study on the use of statistical techniques in the analysis of evolutionary algorithms’ behaviour over optimization problems. A study about the required conditions for statistical analysis of the results is presented by using some models of evolutionary algorithms for real-coding optimization. This study is conducted in two ways: single-problem analysis and multiple-problem analysis. The results obtained state that a parametric statistical analysis could not be appropriate specially when we deal with multiple-problem results. In multiple-problem analysis, we propose the use of non-parametric statistical tests given that they are less restrictive than parametric ones and they can be used over small size samples of results. As a case study, we analyze the published results for the algorithms presented in the CEC’2005 Special Session on Real Parameter Optimization by using non-parametric test procedures. [ABSTRACT FROM AUTHOR]

One important Artificial Intelligence tool for automatic control is the use of fuzzy logic controllers, which are fuzzy rule-based systems comprising expert knowledge in form of linguistic rules. These rules are usually constructed by an expert in the field of interest who can link the facts with the conclusions. However, this way to work sometimes fails to obtain an optimal behaviour. To solve this problem, within the framework of Machine Learning, some Artificial Intelligence techniques could be successfully applied to enhance the controller behaviour. Rule selection methods directly obtain a subset of rules from a given fuzzy rule set, removing inefficient and redundant rules and, thereby, enhancing the controller interpretability, robustness, flexibility and control capability. Besides, different parameter optimization techniques could be applied to improve the system accuracy by inducing a better cooperation among the rules composing the final rule base. This work presents a study of how two new tuning approaches can be applied to improve FLCs obtained from the expert's experience in non trivial problems. Additionally, we analyze the positive synergy between rule selection and tuning techniques as a way to enhance the capability of these methods to obtain more accurate and compact FLCs. Finally, in order to show the good performance of these approaches, we solve a real-world problem for the control of a heating, ventilating and air conditioning system. [ABSTRACT FROM AUTHOR]

Real-world electrical engineering problems can take advantage of the last Data Analysis methodologies. In this paper we will show that Genetic Fuzzy Rule-Based Systems and Genetic Programming techniques are good choices for tackling with some practical modeling problems. We claim that both evolutionary processes may produce good numerical results while providing us with a model that can be interpreted by a human being. We will analyze in detail the characteristics of these two methods and we will compare them to the some of the most popular classical statistical modeling methods and neural networks. [ABSTRACT FROM AUTHOR]

Examines the relationship between seasonal environmental factors and rates of aquatic primary productivity in Terminos Lagoon, Mexico. Influence of surface runoff from fringing mangrove forests on estuarine water column metabolism; Effect of organic carbon on secondary productivity of tropical estuaries; Ability of nutrients exported from mangroves to stimulate phytoplankton production.

An introduction to the special issue of the journal is presented in which the editor discusses various papers discussed during the 23rd International Conference on Industrial, Engineering and Other Applications of Applied Intelligence Systems (IEA/AIE 2010) including real-life applications of artificial intelligence, application of soft computing methods for oil price forecast, and speech recognition of robots.