https://reunir.unir.net/handle/123456789/13559 2024-11-04T15:56:43Z 2024-11-04T15:56:43Z Dey, Nilanjan https://reunir.unir.net/handle/123456789/13588 2022-10-10T11:52:59Z

Editor's Note Dey, Nilanjan The International Journal of Interactive Multimedia and Artificial Intelligence – IJIMAI (ISSN 1989-1660) provides an interdisciplinary forum in which scientists and professionals can share their research results and report new advances in Artificial Intelligence (AI) tools or tools that use AI with interactive multimedia techniques. The present volume (June 2022), consists of 20 articles of diverse applications of great impact in several fields. The issue consistently showcases the utilization of AI techniques or mathematical models with an artificial intelligence base, as a standard element. Different manuscripts on usability and satisfaction, machine learning models, genetic algorithms, computer entertainment technologies, oral pathologies, optimistic motion planning, data analysis for decision making, etc. can be found in this volume. Submitted by Susana Figueroa Navarro (susana.figueroa.n@unir.net) on 2022-10-10T11:52:59Z No. of bitstreams: 1 ijimai_7_4_0.pdf: 69053 bytes, checksum: a695e4b34ddc8592cf4f4ed7c4ad72c5 (MD5); Made available in DSpace on 2022-10-10T11:52:59Z (GMT). No. of bitstreams: 1 ijimai_7_4_0.pdf: 69053 bytes, checksum: a695e4b34ddc8592cf4f4ed7c4ad72c5 (MD5)

Kenye, Lhilo Kala, Rahul https://reunir.unir.net/handle/123456789/13587 2022-10-10T11:49:55Z

Optimistic Motion Planning Using Recursive Sub- Sampling: A New Approach to Sampling-Based Motion Planning Kenye, Lhilo; Kala, Rahul Sampling-based motion planning in the field of robot motion planning has provided an effective approach to finding path for even high dimensional configuration space and with the motivation from the concepts of sampling based-motion planners, this paper presents a new sampling-based planning strategy called Optimistic Motion Planning using Recursive Sub-Sampling (OMPRSS), for finding a path from a source to a destination sanguinely without having to construct a roadmap or a tree. The random sample points are generated recursively and connected by straight lines. Generating sample points is limited to a range and edge connectivity is prioritized based on their distances from the line connecting through the parent samples with the intention to shorten the path. The planner is analysed and compared with some sampling strategies of probabilistic roadmap method (PRM) and the experimental results show agile planning with early convergence. Submitted by Susana Figueroa Navarro (susana.figueroa.n@unir.net) on 2022-10-10T11:49:55Z No. of bitstreams: 1 ijimai_7_4_8.pdf: 762337 bytes, checksum: a07f5f3fddbfa898a9ef7915ac5a0749 (MD5); Made available in DSpace on 2022-10-10T11:49:55Z (GMT). No. of bitstreams: 1 ijimai_7_4_8.pdf: 762337 bytes, checksum: a07f5f3fddbfa898a9ef7915ac5a0749 (MD5)

Khattak, Muhammad Irfan Saleem, Nasir Nawaz, Aamir Ahmed Almani, Aftab Umer, Farhana Verdú, Elena https://reunir.unir.net/handle/123456789/13586 2023-03-17T12:44:23Z

ERBM-SE: Extended Restricted Boltzmann Machine for Multi-Objective Single-Channel Speech Enhancement Khattak, Muhammad Irfan; Saleem, Nasir; Nawaz, Aamir; Ahmed Almani, Aftab; Umer, Farhana; Verdú, Elena Machine learning-based supervised single-channel speech enhancement has achieved considerable research interest over conventional approaches. In this paper, an extended Restricted Boltzmann Machine (RBM) is proposed for the spectral masking-based noisy speech enhancement. In conventional RBM, the acoustic features for the speech enhancement task are layerwise extracted and the feature compression may result in loss of vital information during the network training. In order to exploit the important information in the raw data, an extended RBM is proposed for the acoustic feature representation and speech enhancement. In the proposed RBM, the acoustic features are progressively extracted by multiple-stacked RBMs during the pre-training phase. The hidden acoustic features from the previous RBM are combined with the raw input data that serve as the new inputs to the present RBM. By adding the raw data to RBMs, the layer-wise features related to the raw data are progressively extracted, that is helpful to mine valuable information in the raw data. The results using the TIMIT database showed that the proposed method successfully attenuated the noise and gained improvements in the speech quality and intelligibility. The STOI, PESQ and SDR are improved by 16.86%, 25.01% and 3.84dB over the unprocessed noisy speech. Submitted by Susana Figueroa Navarro (susana.figueroa.n@unir.net) on 2022-10-10T11:41:09Z No. of bitstreams: 1 ijimai_7_4_16.pdf: 1617071 bytes, checksum: 924724aff644ee8bb4ddccaf3ab83aff (MD5); Made available in DSpace on 2022-10-10T11:41:09Z (GMT). No. of bitstreams: 1 ijimai_7_4_16.pdf: 1617071 bytes, checksum: 924724aff644ee8bb4ddccaf3ab83aff (MD5)

Urbina-Nájera, Argelia B. Méndez-Ortega, Luis A. https://reunir.unir.net/handle/123456789/13585 2022-10-10T11:24:45Z

Predictive Model for Taking Decision to Prevent University Dropout Urbina-Nájera, Argelia B.; Méndez-Ortega, Luis A. Dropout is an educational phenomenon studied for decades due to the diversity of its causes, whose effects fall on society's development. This document presents an experimental study to obtain a predictive model that allows anticipating a university dropout. The study uses 51,497 instances with 26 attributes obtained from social sciences, administrative sciences, and engineering collected from 2010 to 2019. Artificial neural networks and decision trees were implemented as classification algorithms, and also, algorithms of attribute selection and resampling methods were used to balance the main class. The results show that the best performing model was that of Random Forest with a Matthew correlation coefficient of 87.43% against 53.39% obtained by artificial neural networks and 94.34% accuracy by Random Forest. The model has allowed predicting an approximate number of possible dropouts per period, contributing to the involved instances in preventing or reducing dropout in higher education. Submitted by Susana Figueroa Navarro (susana.figueroa.n@unir.net) on 2022-10-10T11:24:45Z No. of bitstreams: 1 ijimai_7_4_18.pdf: 902273 bytes, checksum: 5b285eb589a27c9f23853cd8fb2adae9 (MD5); Made available in DSpace on 2022-10-10T11:24:45Z (GMT). No. of bitstreams: 1 ijimai_7_4_18.pdf: 902273 bytes, checksum: 5b285eb589a27c9f23853cd8fb2adae9 (MD5)

Adimoolam, M. Mohan, Senthilkumar A., John Srivastava, Gautam https://reunir.unir.net/handle/123456789/13584 2022-10-10T11:18:19Z

A Novel Technique to Detect and Track Multiple Objects in Dynamic Video Surveillance Systems Adimoolam, M.; Mohan, Senthilkumar; A., John; Srivastava, Gautam Video surveillance is one of the important state of the art systems to be utilized in order to monitor different areas of modern society surveillance like the general public surveillance system, city traffic monitoring system, and forest monitoring system. Hence, surveillance systems have become especially relevant in the digital era. The needs of the video surveillance systems and its video analytics have become inevitable due to an increase in crimes and unethical behavior. Thus enabling the tracking of individuals object in video surveillance is an essential part of modern society. With the advent of video surveillance, performance measures for such surveillance also need to be improved to keep up with the ever increasing crime rates. So far, many methodologies relating to video surveillance have been introduced ranging from single object detection with a single or multiple cameras to multiple object detection using single or multiple cameras. Despite this, performance benchmarks and metrics need further improvements. While mechanisms exist for single or multiple object detection and prediction on videos or images, none can meet the criteria of detection and tracking of multiple objects in static as well as dynamic environments. Thus, real-world multiple object detection and prediction systems need to be introduced that are both accurate as well as fast and can also be adopted in static and dynamic environments. This paper introduces the Densely Feature selection Convolutional neural Network – Hyper Parameter tuning (DFCNHP) and it is a hybrid protocol with faster prediction time and high accuracy levels. The proposed system has successfully tracked multiple objects from multiple channels and is a combination of dense block, feature selection, background subtraction and Bayesian methods. The results of the experiment conducted demonstrated an accuracy of 98% and 1.11 prediction time and these results have also been compared with existing methods such as Kalman Filtering (KF) and Deep Neural Network (DNN). Submitted by Susana Figueroa Navarro (susana.figueroa.n@unir.net) on 2022-10-10T11:18:19Z No. of bitstreams: 1 ijimai_7_4_10.pdf: 990877 bytes, checksum: b9671b8b61f9af6d09f47c0ac796232a (MD5); Made available in DSpace on 2022-10-10T11:18:19Z (GMT). No. of bitstreams: 1 ijimai_7_4_10.pdf: 990877 bytes, checksum: b9671b8b61f9af6d09f47c0ac796232a (MD5)

Anari, Z. Hatamlou, A. Anari, B. https://reunir.unir.net/handle/123456789/13583 2022-10-10T10:55:42Z

Automatic Finding Trapezoidal Membership Functions in Mining Fuzzy Association Rules Based on Learning Automata Anari, Z.; Hatamlou, A.; Anari, B. Association rule mining is an important data mining technique used for discovering relationships among all data items. Membership functions have a significant impact on the outcome of the mining association rules. An important challenge in fuzzy association rule mining is finding an appropriate membership functions, which is an optimization issue. In the most relevant studies of fuzzy association rule mining, only triangle membership functions are considered. This study, as the first attempt, used a team of continuous action-set learning automata (CALA) to find both the appropriate number and positions of trapezoidal membership functions (TMFs). The spreads and centers of the TMFs were taken into account as parameters for the research space and a new approach for the establishment of a CALA team to optimize these parameters was introduced. Additionally, to increase the convergence speed of the proposed approach and remove bad shapes of membership functions, a new heuristic approach has been proposed. Experiments on two real data sets showed that the proposed algorithm improves the efficiency of the extracted rules by finding optimized membership functions. Submitted by Susana Figueroa Navarro (susana.figueroa.n@unir.net) on 2022-10-10T10:55:42Z No. of bitstreams: 1 ijimai_7_4_3.pdf: 2909923 bytes, checksum: 99e7b28cabb8d299891af90b3a8ce380 (MD5); Made available in DSpace on 2022-10-10T10:55:42Z (GMT). No. of bitstreams: 1 ijimai_7_4_3.pdf: 2909923 bytes, checksum: 99e7b28cabb8d299891af90b3a8ce380 (MD5)

Kamel Benamara, Nadir Zigh, Ehlem Boudghene Stambouli, Tarik Keche, Mokhtar https://reunir.unir.net/handle/123456789/13582 2022-10-10T10:48:59Z

Towards a Robust Thermal-Visible Heterogeneous Face Recognition Approach Based on a Cycle Generative Adversarial Network Kamel Benamara, Nadir; Zigh, Ehlem; Boudghene Stambouli, Tarik; Keche, Mokhtar Security is a sensitive area that concerns all authorities around the world due to the emerging terrorism phenomenon. Contactless biometric technologies such as face recognition have grown in interest for their capacity to identify probe subjects without any human interaction. Since traditional face recognition systems use visible spectrum sensors, their performances decrease rapidly when some visible imaging phenomena occur, mainly illumination changes. Unlike the visible spectrum, Infrared spectra are invariant to light changes, which makes them an alternative solution for face recognition. However, in infrared, the textural information is lost. We aim, in this paper, to benefit from visible and thermal spectra by proposing a new heterogeneous face recognition approach. This approach includes four scientific contributions. The first one is the annotation of a thermal face database, which has been shared via Github with all the scientific community. The second is the proposition of a multi-sensors face detector model based on the last YOLO v3 architecture, able to detect simultaneously faces captured in visible and thermal images. The third contribution takes up the challenge of modality gap reduction between visible and thermal spectra, by applying a new structure of CycleGAN, called TV-CycleGAN, which aims to synthesize visible-like face images from thermal face images. This new thermal-visible synthesis method includes all extreme poses and facial expressions in color space. To show the efficacy and the robustness of the proposed TV-CycleGAN, experiments have been applied on three challenging benchmark databases, including different real-world scenarios: TUFTS and its aligned version, NVIE and PUJ. The qualitative evaluation shows that our method generates more realistic faces. The quantitative one demonstrates that the proposed TV -CycleGAN gives the best improvement on face recognition rates. Therefore, instead of applying a direct matching from thermal to visible images which allows a recognition rate of 47,06% for TUFTS Database, a proposed TV-CycleGAN ensures accuracy of 57,56% for the same database. It contributes to a rate enhancement of 29,16%, and 15,71% for NVIE and PUJ databases, respectively. It reaches an accuracy enhancement of 18,5% for the aligned TUFTS database. It also outperforms some recent state of the art methods in terms of F1-Score, AUC/EER and other evaluation metrics. Furthermore, it should be mentioned that the obtained visible synthesized face images using TV-CycleGAN method are very promising for thermal facial landmark detection as a fourth contribution of this paper. Submitted by Susana Figueroa Navarro (susana.figueroa.n@unir.net) on 2022-10-10T10:48:59Z No. of bitstreams: 1 ijimai_7_4_12.pdf: 2334388 bytes, checksum: 8b23aedc9101f894bb60fa7392804bbd (MD5); Made available in DSpace on 2022-10-10T10:48:59Z (GMT). No. of bitstreams: 1 ijimai_7_4_12.pdf: 2334388 bytes, checksum: 8b23aedc9101f894bb60fa7392804bbd (MD5)

Medel, Diego González-González, Carina V. Aciar, Silvana https://reunir.unir.net/handle/123456789/13581 2022-10-10T10:34:17Z

Social Relations and Methods in Recommender Systems: A Systematic Review Medel, Diego; González-González, Carina; V. Aciar, Silvana With the constant growth of information, data sparsity problems, and cold start have become a complex problem in obtaining accurate recommendations. Currently, authors consider the user's historical behavior and find contextual information about the user, such as social relationships, time information, and location. In this work, a systematic review of the literature on recommender systems that use the information on social relationships between users was carried out. As the main findings, social relations were classified into three groups: trust, friend activities, and user interactions. Likewise, the collaborative filtering approach was the most used, and with the best results, considering the methods based on memory and model. The most used metrics that we found, and the recommendation methods studied in mobile applications are presented. The information provided by this study can be valuable to increase the precision of the recommendations. Submitted by Susana Figueroa Navarro (susana.figueroa.n@unir.net) on 2022-10-10T10:34:17Z No. of bitstreams: 1 ijimai_7_4_1.pdf: 420403 bytes, checksum: 2eb5843e3904855c004b7899a5fde684 (MD5); Made available in DSpace on 2022-10-10T10:34:17Z (GMT). No. of bitstreams: 1 ijimai_7_4_1.pdf: 420403 bytes, checksum: 2eb5843e3904855c004b7899a5fde684 (MD5)

Alonso-Misol Gerlache, Héctor Moreno-Ger, Pablo de-la-Fuente-Valentín, Luis https://reunir.unir.net/handle/123456789/13580 2023-03-17T12:43:42Z

Towards the Grade’s Prediction. A Study of Different Machine Learning Approaches to Predict Grades from Student Interaction Data Alonso-Misol Gerlache, Héctor; Moreno-Ger, Pablo; de-la-Fuente-Valentín, Luis There is currently an open problem within the field of Artificial Intelligence applied to the educational field, which is the prediction of students’ grades. This problem aims to predict early school failure and dropout, and to determine the well-founded analysis of student performance for the improvement of educational quality. This document deals the problem of predicting grades of UNIR university master’s degree students in the on-line mode, proposing a working model and comparing different technologies to determine which one fits best with the available data set. In order to make the predictions, the dataset was submitted to a cleaning and analysis phases, being prepared for the use of Machine Learning algorithms, such as Naive Bayes, Decision Tree, Random Forest and Neural Networks. A comparison is made that addresses a double prediction on a homogeneous set of input data, predicting the final grade per subject and the final master’s degree grade. The results were obtained demonstrate that the use of these techniques makes possible the grade predictions. The data gives some figures in which we can see how Artificial Intelligence is able to predict situations with an accuracy above 96%. Submitted by Susana Figueroa Navarro (susana.figueroa.n@unir.net) on 2022-10-10T10:27:59Z No. of bitstreams: 1 ijimai_7_4_17.pdf: 951607 bytes, checksum: 7f940735fa1bb87d7f4622b3cc05daae (MD5); Made available in DSpace on 2022-10-10T10:27:59Z (GMT). No. of bitstreams: 1 ijimai_7_4_17.pdf: 951607 bytes, checksum: 7f940735fa1bb87d7f4622b3cc05daae (MD5)

Alcaide, Asier Patricio, Miguel A. Berlanga, Antonio Arroyo, Angel Cuadrado Gallego, Juan J. https://reunir.unir.net/handle/123456789/13579 2022-10-10T10:21:24Z

LIPSNN: A Light Intrusion-Proving Siamese Neural Network Model for Facial Verification Alcaide, Asier; Patricio, Miguel A.; Berlanga, Antonio; Arroyo, Angel; Cuadrado Gallego, Juan J. Facial verification has experienced a breakthrough in recent years, not only due to the improvement in accuracy of the verification systems but also because of their increased use. One of the main reasons for this has been the appearance and use of new models of Deep Learning to address this problem. This extension in the use of facial verification has had a high impact due to the importance of its applications, especially on security, but the extension of its use could be significantly higher if the problem of the required complex calculations needed by the Deep Learning models, that usually need to be executed on machines with specialised hardware, were solved. That would allow the use of facial verification to be extended, making it possible to run this software on computers with low computing resources, such as Smartphones or tablets. To solve this problem, this paper presents the proposal of a new neural model, called Light Intrusion-Proving Siamese Neural Network, LIPSNN. This new light model, which is based on Siamese Neural Networks, is fully presented from the description of its two block architecture, going through its development, including its training with the well- known dataset Labeled Faces in the Wild, LFW; to its benchmarking with other traditional and deep learning models for facial verification in order to compare its performance for its use in low computing resources systems for facial recognition. For this comparison the attribute parameters, storage, accuracy and precision have been used, and from the results obtained it can be concluded that the LIPSNN can be an alternative to the existing models to solve the facet problem of running facial verification in low computing resource devices. Submitted by Susana Figueroa Navarro (susana.figueroa.n@unir.net) on 2022-10-10T10:21:24Z No. of bitstreams: 1 ijimai_7_4_11.pdf: 666771 bytes, checksum: 2b1ac5da5fc7c530be14580f1b8cf2a2 (MD5); Made available in DSpace on 2022-10-10T10:21:24Z (GMT). No. of bitstreams: 1 ijimai_7_4_11.pdf: 666771 bytes, checksum: 2b1ac5da5fc7c530be14580f1b8cf2a2 (MD5)

Singh Lamba, Puneet Virmani, Deepali Pillai, Manu S. Chaudhary, Gopal https://reunir.unir.net/handle/123456789/13570 2022-10-07T09:48:08Z

Multimodal Human Eye Blink Recognition Using Z-score Based Thresholding and Weighted Features Singh Lamba, Puneet; Virmani, Deepali; Pillai, Manu S.; Chaudhary, Gopal A novel real-time multimodal eye blink detection method using an amalgam of five unique weighted features extracted from the circle boundary formed from the eye landmarks is proposed. The five features, namely (Vertical Head Positioning, Orientation Factor, Proportional Ratio, Area of Intersection, and Upper Eyelid Radius), provide imperative gen (z score threshold) accurately predicting the eye status and thus the blinking status. An accurate and precise algorithm employing the five weighted features is proposed to predict eye status (open/close). One state-of-the-art dataset ZJU (eye-blink), is used to measure the performance of the method. Precision, recall, F1-score, and ROC curve measure the proposed method performance qualitatively and quantitatively. Increased accuracy (of around 97.2%) and precision (97.4%) are obtained compared to other existing unimodal approaches. The efficiency of the proposed method is shown to outperform the state-of-the-art methods. Submitted by Susana Figueroa Navarro (susana.figueroa.n@unir.net) on 2022-10-07T09:48:08Z No. of bitstreams: 1 ijimai_7_4_9.pdf: 1644982 bytes, checksum: aa8f04a311082d578e06a4b2e090bb6f (MD5); Made available in DSpace on 2022-10-07T09:48:08Z (GMT). No. of bitstreams: 1 ijimai_7_4_9.pdf: 1644982 bytes, checksum: aa8f04a311082d578e06a4b2e090bb6f (MD5)

Touza, Ramón Martínez Torres, Javier Álvarez, María Roca, Javier https://reunir.unir.net/handle/123456789/13569 2023-07-18T13:14:05Z

Obtaining Anti-Missile Decoy Launch Solution from a Ship Using Machine Learning Techniques Touza, Ramón; Martínez Torres, Javier; Álvarez, María; Roca, Javier One of the most dangerous situations a warship may face is a missile attack launched from other ships, aircrafts, submarines or land. In addition, given the current scenario, it is not ruled out that a terrorist group may acquire missiles and use them against ships operating close to the coast, which increases their vulnerabilitydue to the limited reaction time. One of the means the ship has for its defense are decoys, designed to deceive the enemy missile. However, for their use to be effective it is necessary to obtain, in a quick way, a valid launching solution. The purpose of this article is to design a methodology to solve the problem of decoy launching and to provide the ship immediately with the necessary data to make the firing decision. To solve the problem machine learning models (neural networks and support vector machines) and a set of training data obtained in simulations will be used. The performance measures obtained with the implementation of multilayer perceptron models allow the replacement of the current procedures based on tables and launching rules with machine learning algorithms that are more flexible and adaptable to a larger number of scenarios. Submitted by Susana Figueroa Navarro (susana.figueroa.n@unir.net) on 2022-10-07T09:37:22Z No. of bitstreams: 1 ijimai_7_4_14.pdf: 1088848 bytes, checksum: 5bcc4cde49dc064dda3c1146ecf9f6e4 (MD5); Made available in DSpace on 2022-10-07T09:37:22Z (GMT). No. of bitstreams: 1 ijimai_7_4_14.pdf: 1088848 bytes, checksum: 5bcc4cde49dc064dda3c1146ecf9f6e4 (MD5)

Laishram, Anuradha Thongam, Khelchandra https://reunir.unir.net/handle/123456789/13568 2022-10-07T09:29:14Z

Automatic Classification of Oral Pathologies Using Orthopantomogram Radiography Images Based on Convolutional Neural Network Laishram, Anuradha; Thongam, Khelchandra An attempt has been made to device a robust method to classify different oral pathologies using Orthopantomogram (OPG) images based on Convolutional Neural Network (CNN). This system will provide a novel approach for the classification of types of teeth (viz., incisors and molar teeth) and also some underlying oral anomalies such as fixed partial denture (cap) and impacted teeth. To this end, various image preprocessing techniques are performed. The input OPG images are resized, pixels are scaled and erroneous data are excluded. The proposed algorithm is implemented using CNN with Dropout and the fully connected layer has been trained using hybrid GA-BP learning. Using the Dropout regularization technique, over fitting has been avoided and thereby making the network to correctly classify the objects. The CNN has been implemented with different convolutional layers and the highest accuracy of 97.92% has been obtained with two convolutional layers. Submitted by Susana Figueroa Navarro (susana.figueroa.n@unir.net) on 2022-10-07T09:29:14Z No. of bitstreams: 1 ijimai_7_4_6.pdf: 928867 bytes, checksum: c993b0942ed94e191524a4b12964d850 (MD5); Made available in DSpace on 2022-10-07T09:29:14Z (GMT). No. of bitstreams: 1 ijimai_7_4_6.pdf: 928867 bytes, checksum: c993b0942ed94e191524a4b12964d850 (MD5)

Novo-Lourés, María Lage, Yeray Pavón, Reyes Laza, Rosalía Ruano-Ordás, David Méndez, José Ramón https://reunir.unir.net/handle/123456789/13567 2022-10-07T08:59:09Z

Improving Pipelining Tools for Pre-processing Data Novo-Lourés, María; Lage, Yeray; Pavón, Reyes; Laza, Rosalía; Ruano-Ordás, David; Méndez, José Ramón The last several years have seen the emergence of data mining and its transformation into a powerful tool that adds value to business and research. Data mining makes it possible to explore and find unseen connections between variables and facts observed in different domains, helping us to better understand reality. The programming methods and frameworks used to analyse data have evolved over time. Currently, the use of pipelining schemes is the most reliable way of analysing data and due to this, several important companies are currently offering this kind of services. Moreover, several frameworks compatible with different programming languages are available for the development of computational pipelines and many research studies have addressed the optimization of data processing speed. However, as this study shows, the presence of early error detection techniques and developer support mechanisms is very limited in these frameworks. In this context, this study introduces different improvements, such as the design of different types of constraints for the early detection of errors, the creation of functions to facilitate debugging of concrete tasks included in a pipeline, the invalidation of erroneous instances and/or the introduction of the burst-processing scheme. Adding these functionalities, we developed Big Data Pipelining for Java (BDP4J, https://github.com/sing-group/bdp4j), a fully functional new pipelining framework that shows the potential of these features. Submitted by Susana Figueroa Navarro (susana.figueroa.n@unir.net) on 2022-10-07T08:59:09Z No. of bitstreams: 1 ijimai_7_4_19.pdf: 796775 bytes, checksum: 2f9a55a24d7cdaefca917b6c0cd3a578 (MD5); Made available in DSpace on 2022-10-07T08:59:09Z (GMT). No. of bitstreams: 1 ijimai_7_4_19.pdf: 796775 bytes, checksum: 2f9a55a24d7cdaefca917b6c0cd3a578 (MD5)

Deore, Mahendra Kulkarni, Uday https://reunir.unir.net/handle/123456789/13566 2022-10-07T08:51:31Z

MDFRCNN: Malware Detection using Faster Region Proposals Convolution Neural Network Deore, Mahendra; Kulkarni, Uday Technological advancement of smart devices has opened up a new trend: Internet of Everything (IoE), where all devices are connected to the web. Large scale networking benefits the community by increasing connectivity and giving control of physical devices. On the other hand, there exists an increased ‘Threat’ of an ‘Attack’. Attackers are targeting these devices, as it may provide an easier ‘backdoor entry to the users’ network’.MALicious softWARE (MalWare) is a major threat to user security. Fast and accurate detection of malware attacks are the sine qua non of IoE, where large scale networking is involved. The paper proposes use of a visualization technique where the disassembled malware code is converted into gray images, as well as use of Image Similarity based Statistical Parameters (ISSP) such as Normalized Cross correlation (NCC), Average difference (AD), Maximum difference (MaxD), Singular Structural Similarity Index Module (SSIM), Laplacian Mean Square Error (LMSE), MSE and PSNR. A vector consisting of gray image with statistical parameters is trained using a Faster Region proposals Convolution Neural Network (F-RCNN) classifier. The experiment results are promising as the proposed method includes ISSP with F-RCNN training. Overall training time of learning the semantics of higher-level malicious behaviors is less. Identification of malware (testing phase) is also performed in less time. The fusion of image and statistical parameter enhances system performance with greater accuracy. The benchmark database from Microsoft Malware Classification challenge has been used to analyze system performance, which is available on the Kaggle website. An overall average classification accuracy of 98.12% is achieved by the proposed method. Submitted by Susana Figueroa Navarro (susana.figueroa.n@unir.net) on 2022-10-07T08:51:31Z No. of bitstreams: 1 ijimai_7_4_13.pdf: 2185179 bytes, checksum: 7e73cb85b1421ad1e8b871a6a5ffd9e9 (MD5); Made available in DSpace on 2022-10-07T08:51:31Z (GMT). No. of bitstreams: 1 ijimai_7_4_13.pdf: 2185179 bytes, checksum: 7e73cb85b1421ad1e8b871a6a5ffd9e9 (MD5)

Ahamed, Jameel Manan Koli, Abdul Ahmad, Khaleel Alam Jamal, Mohd. Gupta, B. B. https://reunir.unir.net/handle/123456789/13565 2022-10-07T08:26:31Z

CDPS-IoT: Cardiovascular Disease Prediction System Based on IoT using Machine Learning Ahamed, Jameel; Manan Koli, Abdul; Ahmad, Khaleel; Alam Jamal, Mohd.; Gupta, B. B. Internet of Things, Machine learning, and Cloud computing are the emerging domains of information communication and technology. These techniques can help to save the life of millions in the medical assisted environment and can be utilized in health-care system where health expertise is less available. Fast food consumption increased from the past few decades, which makes up cholesterol, diabetes, and many more problems that affect the heart and other organs of the body. Changing lifestyle is another parameter that results in health issues including cardio-vascular diseases. Affirming to the World Health Organization, the cardiovascular diseases, or heart diseases lead to more death than any other disease globally. The objective of this research is to analyze the available data pertaining to cardiovascular diseases for prediction of heart diseases at an earlier stage to prevent it from occurring. The dataset of heart disease patients was taken from Jammu and Kashmir, India and stored over the cloud. Stored data is then pre-processed and further analyzed using machine learning techniques for the prediction of heart diseases. The analysis of the dataset using numerous machines learning techniques like Random Forest, Decision Tree, Naive based, K-nearest neighbors, and Support Vector Machine revealed the performance metrics (F1 Score, Precision and Recall) for all the techniques which shows that Naive Bayes is better without parameter tuning while Random Forest algorithm proved as the best technique with hyperparameter tuning. In this paper, the proposed model is developed in such a systematic way that the clinical data can be obtained through the use of IoT with the help of available medical sensors to predict cardiovascular diseases on a real-time basis. Submitted by Susana Figueroa Navarro (susana.figueroa.n@unir.net) on 2022-10-07T08:26:31Z No. of bitstreams: 1 ijimai_7_4_7.pdf: 503531 bytes, checksum: 7e4b62c1a9e3651de585e8e4a5703f64 (MD5); Made available in DSpace on 2022-10-07T08:26:31Z (GMT). No. of bitstreams: 1 ijimai_7_4_7.pdf: 503531 bytes, checksum: 7e4b62c1a9e3651de585e8e4a5703f64 (MD5)

Bobadilla, Jesús Gutiérrez, Abraham Alonso, Santiago González-Prieto, Ángel https://reunir.unir.net/handle/123456789/13564 2022-10-07T08:19:01Z

Neural Collaborative Filtering Classification Model to Obtain Prediction Reliabilities Bobadilla, Jesús; Gutiérrez, Abraham; Alonso, Santiago; González-Prieto, Ángel Neural collaborative filtering is the state of art field in the recommender systems area; it provides some models that obtain accurate predictions and recommendations. These models are regression-based, and they just return rating predictions. This paper proposes the use of a classification-based approach, returning both rating predictions and their reliabilities. The extra information (prediction reliabilities) can be used in a variety of relevant collaborative filtering areas such as detection of shilling attacks, recommendations explanation or navigational tools to show users and items dependences. Additionally, recommendation reliabilities can be gracefully provided to users: “probably you will like this film”, “almost certainly you will like this song”, etc. This paper provides the proposed neural architecture; it also tests that the quality of its recommendation results is as good as the state of art baselines. Remarkably, individual rating predictions are improved by using the proposed architecture compared to baselines. Experiments have been performed making use of four popular public datasets, showing generalizable quality results. Overall, the proposed architecture improves individual rating predictions quality, maintains recommendation results and opens the doors to a set of relevant collaborative filtering fields. Submitted by Susana Figueroa Navarro (susana.figueroa.n@unir.net) on 2022-10-07T08:19:01Z No. of bitstreams: 1 ijimai_7_4_2.pdf: 812772 bytes, checksum: 1d0ba7a63f430bcaefa94dd684e0e3d7 (MD5); Made available in DSpace on 2022-10-07T08:19:01Z (GMT). No. of bitstreams: 1 ijimai_7_4_2.pdf: 812772 bytes, checksum: 1d0ba7a63f430bcaefa94dd684e0e3d7 (MD5)

Song, Yang https://reunir.unir.net/handle/123456789/13563 2022-10-07T08:08:16Z

Research on the Application of Computer Graphic Advertisement Design Based on a Genetic Algorithm and TRIZ Theory Song, Yang In view of the shortcomings of the traditional thinking of computer graphic advertising design, this paper introduces TRIZ innovative thinking to design computer advertising. First of all, combined with specific cases of computer creative print advertising, this paper analyzes the creative methods of stimulating divergent thinking, aggregation thinking and transformation thinking from the innovation principle of TRIZ theory as the origin, and applies them to the creative mechanism and application program of print advertising creativity. The whole process is led by rational principles of perceptual thinking, driven by specific principles of abstract imagination, to explore the thinking source of creative design essence of print advertising. The theory and its application mechanism become a new thinking method and application attempt in the creative field of print advertisement. Then, based on the TRIZ innovation theory, the business model of advertising content arrangement is constructed, and the mathematical model is constructed according to the planning business media resource planning on the business model to realize the multi-objective optimization of efficient use of orders and precise delivery of time. Finally, a multi-objective optimization mathematical model of parallel genetic algorithm is designed to solve the advertisement content arrangement. The innovative thinking of TRIZ and the application of genetic algorithm in content arrangement of computer graphic advertisement design are verified by experiments. Submitted by Susana Figueroa Navarro (susana.figueroa.n@unir.net) on 2022-10-07T08:08:16Z No. of bitstreams: 1 ijimai_7_4_4.pdf: 668673 bytes, checksum: 98d196361d54488f275d18da8a88f9a8 (MD5); Made available in DSpace on 2022-10-07T08:08:16Z (GMT). No. of bitstreams: 1 ijimai_7_4_4.pdf: 668673 bytes, checksum: 98d196361d54488f275d18da8a88f9a8 (MD5)

Fernández-García, Antonio Jesús Preciado, Juan Carlos Prieto, Álvaro E. Sánchez-Figueroa, Fernando Gutiérrez, Juan D. https://reunir.unir.net/handle/123456789/13562 2023-03-17T12:42:30Z

CompareML: A Novel Approach to Supporting Preliminary Data Analysis Decision Making Fernández-García, Antonio Jesús; Preciado, Juan Carlos; Prieto, Álvaro E.; Sánchez-Figueroa, Fernando; Gutiérrez, Juan D. There are a large number of machine learning algorithms as well as a wide range of libraries and services that allow one to create predictive models. With machine learning and artificial intelligence playing a major role in dealing with engineering problems, practising engineers often come to the machine learning field so overwhelmed with the multitude of possibilities that they find themselves needing to address difficulties before actually starting on carrying out any work. Datasets have intrinsic properties that make it hard to select the algorithm that is best suited to some specific objective, and the ever-increasing number of providers together make this selection even harder. These were the reasons underlying the design of CompareML, an approach to supporting the evaluation and comparison of machine learning libraries and services without deep machine learning knowledge. CompareML makes it easy to compare the performance of different models by using well-known classification and regression algorithms already made available by some of the most widely used providers. It facilitates the practical application of methods and techniques of artificial intelligence that let a practising engineer decide whether they might be used to resolve hitherto intractable problems. Thus, researchers and engineering practitioners can uncover the potential of their datasets for the inference of new knowledge by selecting the most appropriate machine learning algorithm and determining the provider best suited to their data. Submitted by Susana Figueroa Navarro (susana.figueroa.n@unir.net) on 2022-10-07T08:02:37Z No. of bitstreams: 1 ijimai_7_4_20.pdf: 1548182 bytes, checksum: 3c51065ee7e9093b4a0eeba4466c21da (MD5); Made available in DSpace on 2022-10-07T08:02:37Z (GMT). No. of bitstreams: 1 ijimai_7_4_20.pdf: 1548182 bytes, checksum: 3c51065ee7e9093b4a0eeba4466c21da (MD5)

López Ibáñez, Manuel Romero-Hernández, Alejandro Manero, Borja Guijarro, María https://reunir.unir.net/handle/123456789/13561 2022-10-07T07:48:35Z

Computer Entertainment Technologies for the Visually Impaired: An Overview López Ibáñez, Manuel; Romero-Hernández, Alejandro; Manero, Borja; Guijarro, María Over the last years, works related to accessible technologies have increased both in number and in quality. This work presents a series of articles which explore different trends in the field of accessible video games for the blind or visually impaired. Reviewed articles are distributed in four categories covering the following subjects: (1) video game design and architecture, (2) video game adaptations, (3) accessible games as learning tools or treatments and (4) navigation and interaction in virtual environments. Current trends in accessible game design are also analysed, and data is presented regarding keyword use and thematic evolution over time. As a conclusion, a relative stagnation in the field of human-computer interaction for the blind is detected. However, as the video game industry is becoming increasingly interested in accessibility, new research opportunities are starting to appear. Submitted by Susana Figueroa Navarro (susana.figueroa.n@unir.net) on 2022-10-07T07:48:35Z No. of bitstreams: 1 ijimai_7_4_5.pdf: 611853 bytes, checksum: de5b9d70e405c0656b23bed7fe0e6913 (MD5); Made available in DSpace on 2022-10-07T07:48:35Z (GMT). No. of bitstreams: 1 ijimai_7_4_5.pdf: 611853 bytes, checksum: de5b9d70e405c0656b23bed7fe0e6913 (MD5)

Diaz, Moises Crispo, Gioele Parziale, Antonio Marcelli, Angelo Ferrer, Miguel A. https://reunir.unir.net/handle/123456789/13560 2022-10-07T07:52:51Z

Writing Order Recovery in Complex and Long Static Handwriting Diaz, Moises; Crispo, Gioele; Parziale, Antonio; Marcelli, Angelo; Ferrer, Miguel A. The order in which the trajectory is executed is a powerful source of information for recognizers. However, there is still no general approach for recovering the trajectory of complex and long handwriting from static images. Complex specimens can result in multiple pen-downs and in a high number of trajectory crossings yielding agglomerations of pixels (also known as clusters). While the scientific literature describes a wide range of approaches for recovering the writing order in handwriting, these approaches nevertheless lack a common evaluation metric. In this paper, we introduce a new system to estimate the order recovery of thinned static trajectories, which allows to effectively resolve the clusters and select the order of the executed pendowns. We evaluate how knowing the starting points of the pen-downs affects the quality of the recovered writing. Once the stability and sensitivity of the system is analyzed, we describe a series of experiments with three publicly available databases, showing competitive results in all cases. We expect the proposed system, whose code is made publicly available to the research community, to reduce potential confusion when the order of complex trajectories are recovered, and this will in turn make the trajectories recovered to be viable for further applications, such as velocity estimation. Submitted by Susana Figueroa Navarro (susana.figueroa.n@unir.net) on 2022-10-07T07:33:24Z No. of bitstreams: 1 ijimai_7_4_15.pdf: 1326471 bytes, checksum: f521c2d83796761e8152f7b955502258 (MD5); Made available in DSpace on 2022-10-07T07:33:24Z (GMT). No. of bitstreams: 1 ijimai_7_4_15.pdf: 1326471 bytes, checksum: f521c2d83796761e8152f7b955502258 (MD5)