Inteligencia Artificial 2022-10-29T02:42:56+02:00 Editor Open Journal Systems <p style="text-align: justify;"><span style="color: #000000;"><strong><em><a style="color: #003366; text-decoration: underline;" href="" target="_blank" rel="noopener">Inteligencia Artificial</a></em></strong><span id="result_box" class="" lang="en"> is an international open access journal promoted by <span class="">the Iberoamerican Society of</span> Artificial Intelligence (<a href="">IBERAMIA</a>). </span></span>Since 1997, the journal publishes high-quality original papers reporting theoretical or applied advances in all areas of Artificial Intelligence. <span style="color: rgba(0, 0, 0, 0.87); font-family: 'Noto Sans', -apple-system, BlinkMacSystemFont, 'Segoe UI', Roboto, Oxygen-Sans, Ubuntu, Cantarell, 'Helvetica Neue', sans-serif; font-size: 14px; font-style: normal; font-variant-ligatures: normal; font-variant-caps: normal; font-weight: 400; letter-spacing: normal; orphans: 2; text-align: justify; text-indent: 0px; text-transform: none; white-space: normal; widows: 2; word-spacing: 0px; -webkit-text-stroke-width: 0px; background-color: #ffffff; text-decoration-thickness: initial; text-decoration-style: initial; text-decoration-color: initial; display: inline !important; float: none;">There are no fees for subscription, publication nor editing tasks<span class="VIiyi" lang="en"><span class="JLqJ4b ChMk0b" data-language-for-alternatives="en" data-language-to-translate-into="es" data-phrase-index="0">.</span></span> <span class="VIiyi" lang="en"><span class="JLqJ4b ChMk0b" data-language-for-alternatives="en" data-language-to-translate-into="es" data-phrase-index="0">Articles can be written in English, Spanish or Portuguese and <a href="">will be subjected</a> to a double-blind peer review process.</span></span> <span class="VIiyi" lang="en"><span class="JLqJ4b ChMk0b" data-language-for-alternatives="en" data-language-to-translate-into="es" data-phrase-index="0">The journal is abstracted and indexed in several <a href="">data bases</a>.</span></span><br /></span></p> Ontology-Based Traffic Accident Information Extraction on Twitter In Indonesia 2022-02-25T02:58:39+01:00 Nur Aini Rakhmawati Yasin Awwab Ahmad Choirun Najib Ahmad Irsyad <p>Traffic accidents become one of the events that often occur in Indonesia. From the three-monthly report by the Indonesian National Police Traffic Police, there are about 25,000 traffic accidents. Many social media users, especially Twitter, share information about traffic accidents. Twitter has various information regarding traffic accidents. Therefore, this study aims to process and map information about traffic accidents contained on Twitter in Indonesia language.&nbsp; We use the domain ontology and Named-Entity Recognition for the data extraction process. Named-Entity Recognition is used for obtaining keywords from a tweet based on class categories such as actor, time, location, and information on the cause of the accident. This research generates a Named Entity Recognition (NER) model that can provide a reasonably accurate level of accuracy. Also, we create an ontology that can categorize the causes of traffic accidents based on the Directorate General of the Land Transportation Office, Indonesia. We found that the traffic accidents are generally caused by inadequate vehicle conditions with the main problem in the vehicle caused by brake failure, while environmental factors rarely cause traffic accidents. Moreover, the vehicle is the subclass that mostly appears in the tweets, where car is the most popular actor, followed by truck and motorcycle.</p> 2022-09-25T00:00:00+02:00 Copyright (c) 2022 Iberamia & The Authors A Machine Vision Approach for Recognizing Coastal Fish 2022-07-11T23:07:56+02:00 Afiq Raihan Israt Sharmin B M Marjan Khan Md. Ismail Jabiullah Md. Tarek Habib <p>Coastal fish is one of the prominent marine resources, which takes a necessary role in the economic growth of a country. Because of environmental issues along with other reasons, not only most of the marine resources are diminishing but also many coastal fishes are getting extinct gradually. As a result, the young peoples have insufficient knowledge of coastal fish. This issue can be solved with the use of vision-based technologies. To deal with this situation, a coastal fish recognition system based on machine vision is conceived, which can be approached by the images of coastal fish that are captured with a portable device and identify the fish to recognize fish. Numerous experimental analyses are executed to exhibit the benefit of this proposed expert system. In the beginning, conversion of a color image into a gray-scale image occurs and the gray-scale histogram is developed. Using the histogram-based method, image segmentation is conducted. After that, a set of thirteen features comprising of four classes is extracted to be fed to a classifier. For reducing the number of features, PCA is applied. To recognize coastal fish, three cutting-edge classifiers are performed, where k-NN provides a potential accuracy of up to 98.7%.</p> 2022-09-25T00:00:00+02:00 Copyright (c) 2022 Iberamia & The Authors Feature Selection: Binary Harris Hawk Optimizer Based Biomedical Datasets 2022-08-06T00:38:44+02:00 Hadeel Tariq Ibrahim Wamidh Jalil Mazher Enas Mahmood Jassim <p>Feature selection (FS) is an essential preprocessing step in utmost solutions for the high-dimensional problem to reduce the number of features by deleting irrelevant and redundant data that preserve a suitable grade of classification accuracy. Feature selection can be treated as an optimization problem. Heuristic optimization algorithms are hopeful approaches to solve feature selection problems because of their difficulty, especially in high-dimensional data. Binary Harris hawk optimization (BHHO) is one of the lately suggested metaheuristic algorithms that has been demonstrated to be used more efficiently in facing some optimization problems. Support vector machines (SVMs) are a vital technique that are employed competently to resolve classification issues. We modified the BHHO algorithm with SVM classifier to solve the feature selection issue. This study suggests BHHO-FS to fix the feature selection problem in biomedical datasets. We ran the proposed approach BHHO-FS on real biomedical datasets with 17 types of cancer for Iraqi patients in 2010-2012. The experimental results demonstrate the supremacy of the proposed BHHO-FS in terms of three performance metrics: feature selection accuracy, runtime and number of selected features compared to four other state-of-art algorithms: Fire Fly (FF) algorithm, Genetic Algorithm (GA), Grasshopper Optimization Algorithm (GOA) and Particle Swarm Algorithm (PSO). Comparative experiments designate the importance of the proposed approach in comparison with the other four mentioned algorithms. The implementation of the proposed BHHO-FS approach on 17 datasets for different types of cancers reveals 99.967% average accuracy.</p> 2022-11-09T00:00:00+01:00 Copyright (c) 2022 Iberamia & The Authors Fast segmentation of point clouds using a convolutional neural network for helping visually impaired people find the closest traversable region 2022-09-16T00:05:40+02:00 Paúl Tinizaray Wilbert Aguilar José Lucio <p>In this paper, we introduce an approach for helping visually impaired people to find the closest-to-user traversable region. The aim of our work is to reduce the computational cost of this task. For this purpose, we develop a convolutional neural network that classifies patches to segment floor regions in a point cloud. Segmented regions are evaluated by their size and position in the point cloud to identify the closest-to-user traversable region. We evaluate our approach using the NYU-v2 dataset and find that by searching only in the lower section of the point cloud, it is possible to reduce the processing time while finding the closest floor regions. Our approach reports a better processing time than related works, making it suitable to quickly find the closest-to-user traversable region in point clouds.</p> 2022-11-23T00:00:00+01:00 Copyright (c) 2022 Iberamia & The Authors Defective sewing stitch semantic segmentation using DeeplabV3+ and EfficientNet 2022-10-18T16:59:00+02:00 Quoc Toan Nguyen <p>Defective stitch inspection is an essential part of garment manufacturing quality assurance. Traditional mechanical defect detection systems are effective, but they are usually customized with handcrafted features that must be operated by a human. Deep learning approaches have recently demonstrated exceptional performance in a wide range of computer vision applications. The requirement for precise detail evaluation, combined with the small size of the patterns, undoubtedly increases the difficulty of identification. Therefore, image segmentation (semantic segmentation) was employed for this task. It is identified as a vital research topic in the field of computer vision, being indispensable in a wide range of real-world applications. Semantic segmentation is a method of labeling each pixel in an image. This is in direct contrast to classification, which assigns a single label to the entire image. And multiple objects of the same class are defined as a single entity. DeepLabV3+ architecture, with encoder-decoder architecture, is the proposed technique. EfficientNet models (B0-B2) were applied as encoders for experimental processes. The encoder is utilized to encode feature maps from the input image. The encoder's significant information is used by the decoder for upsampling and reconstruction of output. Finally, the best model is DeeplabV3+ with EfficientNetB1 which can classify segmented defective sewing stitches with superior performance (<em>MeanIoU</em>: 94.14%).</p> 2022-11-24T00:00:00+01:00 Copyright (c) 2022 Iberamia & The Authors Autonomous UAV object Avoidance with Floyd-warshall differential evolution approach 2022-06-10T13:43:16+02:00 guruprasad YK NageswaraGuptha M N <p>Unmanned Aerial Vehicles (UAVs) are recently focused with significant research attention from commercial to military industries. Due to its wide range of applications such as traffic monitoring, surveillance, aerial photograph and rescue mission, many research studies were conducted related to UAV development. UAV are commonly called as ‘drones’ used to suit dull, dangerous and dirty missions that can be suited by manned aircraft. UAV can be controlled either remotely or using automation approaches so that it can be travelled into predefined path. To make the autonomous UAV, the most complex issue that is faced by UAV is obstacle / object avoidance. Obstacle detection and avoidance are important for UAV and it is the complex problem to solve due to the payload restriction. This will limit the sensor count mounted on the vehicle. Radar was used to find the distance between the object and vehicle. This can help to detect and track the moving objects speed and direction towards the vehicle. This paper considered the object avoidance problem as path planning problem. There were many path planning methods related to UAV which formulates the path planning as an optimization problem to avoid the obstacles. With the consideration, this paper proposed an efficient and optimal approach called Floyd Warshall- Differential evolution (FWDE) approach to detect the frontal obstacles of UAV. Finally, statistical analysis of the simulated environment reveals that the proposed evolutionary method can efficiently avoid both static and dynamic objects for UAVs. This efficient avoidance algorithm for UAV can be experimented with simulation environment with three kinds of scenarios having different number of cells. The obtained accuracy and recall value of the proposed system is 95.21% and 91.56%.</p> 2022-12-05T00:00:00+01:00 Copyright (c) 2022 Iberamia & The Authors A Large-Scale Study of Activation Functions in Modern Deep Neural Network Architectures for Efficient Convergence 2022-10-29T02:42:56+02:00 Andrinandrasana David Rasamoelina Ivan Cík Peter Sincak Marián Mach Lukáš Hruška <p><span dir="ltr" role="presentation">Activation functions play an important role in the convergence of learning algorithms based on neural networks. They</span><br role="presentation"><span dir="ltr" role="presentation">provide neural networks with nonlinear ability and the possibility to fit in any complex data. However, no deep study exists in the</span><br role="presentation"><span dir="ltr" role="presentation">literature on the comportment of activation functions in modern architecture. Therefore, in this research, we compare the 18 most </span><span dir="ltr" role="presentation">used activation functions on multiple datasets (CIFAR-10, CIFAR-100, CALTECH-256) using 4 different models (EfficientNet,</span><br role="presentation"><span dir="ltr" role="presentation">ResNet, a variation of ResNet using the bag of tricks, and MobileNet V3). Furthermore, we explore the shape of the loss</span><br role="presentation"><span dir="ltr" role="presentation">landscape of those different architectures with various activation functions. Lastly, based on the result of our experimentation,</span><br role="presentation"><span dir="ltr" role="presentation">we introduce a new locally quadratic activation function namely Hytana alongside one variation Parametric Hytana which</span><br role="presentation"><span dir="ltr" role="presentation">outperforms common activation functions and address the dying ReLU problem.</span></p> 2022-12-08T00:00:00+01:00 Copyright (c) 2022 Iberamia & The Authors Performance of Deep Learning models with transfer learning for multiple-step-ahead forecasts in monthly time series 2022-09-13T17:34:39+02:00 Martín Solís Luis-Alexander Calvo-Valverde <p>Deep Learning and transfer learning models are being used to generate time series forecasts; however, there is scarce evidence about their performance prediction that it is more evident for monthly time series. The purpose of this paper is to compare Deep Learning models with transfer learning and without transfer learning and other traditional methods used for monthly forecasts to answer three questions about the suitability of Deep Learning and Transfer Learning to generate predictions of time series. Time series of M4 and M3 competitions were used for the experiments. The results suggest that deep learning models based on TCN, LSTM, and CNN with transfer learning tend to surpass the performance prediction of other traditional methods. On the other hand, TCN and LSTM, trained directly on the target time series, got similar or better performance than traditional methods for some forecast horizons.</p> 2022-12-07T00:00:00+01:00 Copyright (c) 2022 Iberamia & The Authors Trends in Research on Artificial Intelligence in Anesthesia: A VOSviewer -Based Bibliometric Analysis 2022-08-19T10:49:11+02:00 Marco Cascella Francesco Perri Alessandro Ottaiano Arturo Cuomo Stefan Wirz Sergio Coluccia <p>Background: The scientific literature on Artificial Intelligence (AI) in anesthesia is rapidly growing. Considering that applications of AI strategies can offer paramount support in clinical decision processes<em>, </em>it is crucial to delineate the research features. Bibliometric analyses can provide an overview of research tendencies useful for supplementary investigations in a research field. Methods: The comprehensive literature about AI in anesthesia was checked in the Web of Science (WOS) core collection. Year of publication, journal metrics including impact factor and quartile, title, document type, topic, and article metric (citations) were extracted. The software tool VOSviewer (version 1.6.17) was implemented for the co-occurrence of keywords and the co-citation analyses, and for evaluating research networks (countries and institutions). Results: Altogether, 288 documents were retrieved from the WOS and 154 articles were included in the analysis. The number of articles increased from 4 articles in 2017 to 37 in 2021. Only 34 were observational investigations and 7 RCTs. The most relevant topic is “anesthesia management”. The research network for countries and institutions shows severe gaps. Conclusion: Research on AI in anesthesia is rapidly developing. Further clinical studies are needed. Although different topics are addressed, scientific collaborations must be implemented.</p> 2022-12-29T00:00:00+01:00 Copyright (c) 2022 Iberamia & The Authors Towards Misbehavior Intelligent Detection Using Guided Machine Learning in Vehicular Ad-hoc Networks (VANET) 2022-08-30T01:17:08+02:00 Omessaad Slama Bechir Alaya Salah Zidi <p class="western" align="justify"><span style="font-size: small;"><span style="vertical-align: inherit;"><span style="vertical-align: inherit;">C-ITS (Cooperative Intelligent Transport Systems) est une nouvelle technologie qui contribue à la réduction des accidents de la circulation et à l'amélioration de la sécurité routière. </span><span style="vertical-align: inherit;">VANET ( </span></span><span style="vertical-align: inherit;"><span style="vertical-align: inherit;">V </span></span><span style="vertical-align: inherit;"><span style="vertical-align: inherit;">ehicular </span></span><span style="vertical-align: inherit;"><span style="vertical-align: inherit;">A </span></span><span style="vertical-align: inherit;"><span style="vertical-align: inherit;">d hoc </span></span><span style="vertical-align: inherit;"><span style="vertical-align: inherit;">N</span></span><span style="vertical-align: inherit;"><span style="vertical-align: inherit;">etworks) sont un système STI basé sur la communication inter-véhicules par la transmission de messages de sécurité de base (BSM), qui sont vulnérables à une variété de comportements inappropriés. </span><span style="vertical-align: inherit;">Pour résoudre ce défi, nous avons développé dans cet article un système de détection de mauvais comportement (MDS) basé sur une approche d'apprentissage automatisé pour identifier et catégoriser les messages de mauvais comportement délivrés par un véhicule sur les VANET à l'aide de la base de données d'extension VeReMi. </span><span style="vertical-align: inherit;">Cette étude examine différents types de classification : dans la classification binaire, toutes sortes d'inconduites ont été regroupées en une seule catégorie « inconduite » ; </span><span style="vertical-align: inherit;">cependant, dans la classification multi-classes pour trois classes, la mauvaise conduite a été divisée en deux classes : les attaques et les fautes. </span><span style="vertical-align: inherit;">Le classificateur a des problèmes substantiels lors de l'apprentissage à partir de données déséquilibrées </span></span><span style="vertical-align: inherit;"><span style="vertical-align: inherit;">w</span></span><span style="vertical-align: inherit;"><span style="vertical-align: inherit;">Lorsque vous travaillez avec des problèmes multi-classes, cela devient considérablement plus complexe. </span><span style="vertical-align: inherit;">Les relations entre les catégories ne sont plus bien définies et il est facile de perdre en efficacité dans une classe tout en s'améliorant dans une autre. </span><span style="vertical-align: inherit;">En conséquence, les résultats ne sont pas cohérents dans l'approche d'apprentissage classique pour la classification multi-classes lors de la classification des comportements répréhensibles dans différents types de classes de comportements répréhensibles. </span><span style="vertical-align: inherit;">Pour résoudre ce problème, nous avons développé une approche nouvelle et puissante appelée "Approche d'apprentissage guidé pour la classification multi-classes" pour réduire le nombre de classes en combinant des comportements inappropriés comparables en un seul. </span><span style="vertical-align: inherit;">Selon les résultats, le classificateur Random Forest surpasse les autres classificateurs.</span></span></span></p> 2022-12-31T00:00:00+01:00 Copyright (c) 2022 Iberamia & The Authors