Telematics and Computing

Agriculture has been an essential and foundational activity for human societies since the dawn of civilization and nowadays serves as the backbone of economies worldwide. Efforts to understand and to enhance agricultural productivity are crucial for addressing global challenges and achieving sustainable development goals. This research study focuses on analyzing the factors influencing production of the flagship crops of the 32 states in Mexico. A regression tree model was employed as an explainable artificial intelligence technique to gain insights into the production patterns. The study utilized a dataset containing various agricultural variables, including territorial extension, precipitation mean, and temperature measurements across different months. Quantitative and qualitative approaches were employed to understand the significance of predictors. Through permutation importance analysis, it was identified that territorial extension, precipitation mean, and specific temperature measures, such as minimum temperature in January and mean temperature in November, had a substantial impact on crop production. Additionally, a visual analysis of the pruned regression tree further confirmed the importance of these predictors. The findings led to the formulation of seven production rules, which provide valuable guidance for agricultural decision-making. The results highlight the potential of the regression tree model as an explainable tool for understanding and predicting crop production.

The number of Additions Per Output Sample (APOS) is, currently, a standard way to estimate the computational complexity of Cascaded Integrator-Comb (CIC) decimators. This metric originates from a perspective with a high level of abstraction, where the amount of additions performed by the CIC hardware architecture accounts in general for the switching activity of the system, and therefore represents the power consumption in a direct proportion. In this paper we introduce an approach that leads towards a more appropriate way to estimate the computational complexity of CIC decimators, which considers the width of the internal buses of the architecture. We employ the pruning scheme by Hogenauer, and we provide explicit formulas to find, through a simple procedure, the number of Atomic Additions Per Output Sample (AAPOS), which accounts for the computational complexity on a bit-by-bit (i.e., atomic) basis. Three detailed examples are included to show how the AAPOS provides different results for systems with apparently equal computational complexity when estimated with APOS, leading to a more precise and trustable estimation.

In this work, the objective is to implement a convolutional neural network (CNN) to automate some processes in companies and with this, avoid human participation and not spend more resources than necessary, in addition to recovering in less time the investments that companies made. The CNN was implemented through the YOLOv5x algorithm, and this classifies 10 different object classes. The corresponding images for the training of the CNN were obtained with the AllImage extension which allowed downloading all the images that were on the screen. In addition to this, multiple dates bank were accessed and large number of images were downloaded having to filter a total of 700,000 images approximately, additionally photos were taken manually to complement the dataset, after the image filter, a dataset with a volume of 10673 images was obtained. Once the image filter was carried out, the network was trained with a single class, applying a k-fold cross-validation to calibrate the model and obtain an optimal percentage of images for training and validation. After having obtained these values, different versions of YOLO (YOLOv3-SPP, YOLOv5x and YOLOv7x) were trained to determine which of them was more accurate, YOLOv5x being the best with an AUC of .8811. Subsequently, the network was trained with the 10 classes. To measure the efficiency and quality of the proposed method, the following metrics were used with their respective values: 95.1% Accuracy, 92.8% Recall, 97.4% Specificity, 97.8% Precision, 95.1% F1 Score. The results obtained show a better classification performance in the neural network with a detection greater than 95% of the evaluation metrics for each class.

This paper presents a digital hardware architecture for a compressed sensing encoder that outperforms the state-of-the-art solution in terms of speed and power metrics, with an acceptable charge in the hardware cost. The proposed core can achieve, on average, an increase of 94% in the maximum sample rate for input data and a reduction of dynamic power consumption of 30% with just 18% of extra hardware utilization, as validated with post-place-and-route information from FPGA-based implementations.

The reliable detection of a seismic even is still an open task due to the complexity of the phenomenon. In addition, the same event is measured with quite different characteristics depending on the site of monitoring, partly due to the seismic signal being mainly the superposition of waves that propagate in multiple directions from their hypocenter, undergoing different attenuation phenomena, reflection, etc. The study of seismic detection has become relevant in the construction of early warning systems in cities. In this work, the application of artificial neural networks for the recognition of records from seismic and non-seismic events is presented. The seismograms and the non-seismic records were used to construct the spectrogram, treated as an RGB image and become the input data for neural networks training-test processes. Three well-known neural networks architectures were used in this study: “ALEXNET, RESNET18, and VGG16”. They are based on convolutional neural networks and were specially constructed for image recognition. Given the characteristics of the images used, and the size of the dataset available, our results show that the three architectures are highly recommended for the proposed application. The data reported for the neural architectures show that it is possible to recognize the frequency elements that conform to a seismic signal. For its final implementation in real-time automatic seismic monitoring for early warning systems, it is proposed to evaluate which of the three architectures consumes less computational resources and presents less response time.

Fisher’s information measure (FIM) allows to study the complexities associated to random signals and systems and has been used in the literature to study EEG and other physiological signals. In this paper, various time-domain definitions of Fisher’s information are extended to the wavelet domain and closed-form expressions for each definition are obtained for fractal signals of parameter $$\alpha $$ α . Fisher information planes are computed in a range of $$\alpha $$ α and based on these, characteristics, properties, and the effect of signal length is also identified. Moreover, based on this, a complete characterization of fractals by wavelet Fisher’s information is presented and the potential application of each definition in practical fractal signal analysis is also highlighted.

The Internet of Things is a new technology in development because its manufacture and consumption in objects as simple as light bulbs and sensors are becoming widespread. However, due to the need to reduce their cost and be affordable to users, most devices have stopped working on security. The previous has motivated the subsequent investigation to find intermediate solutions to protect information security, improving cybersecurity. Therefore, the objective of the present investigation was to experiment on an Internet of Things platform with commercial devices, following the standard settings, to find the risks and be able to propose a solution. The methodology was to build the IoT network, perform ARP and DNS attacks to obtain sensitive information, then add a firewall to prevent the attack. The vulnerabilities could be found using ESP8266, a Wi-Fi spotlight, a Raspberry in the specialized hardware and software for attacks, and the solution with Raspberry as AP. The contribution is to have shown that cybersecurity technologies can and should be considered more formally; on the contrary, if we do not minimize the risk, we could suffer repercussions.

The abstract Multimedia technology plays a crucial role in the development of smart cities, integrating multimedia systems such as sensors, cameras, mobile applications, augmented reality, along with cloud computing, cybersecurity, big data, and the Internet of Things. This combination allows for improving the effectiveness of public services, citizen participation, and the overall quality of life in society. The purpose of this research was to identify through a systematic literature review, the benefits, challenges and issues related to multimedia technology in the construction of smart cities. The consulted sources of information included scientific articles, research projects, case studies, and electronic books. Among the results, the key opportunities for urban development exposed through the effective adoption of digital technologies, as well as innovative solutions that promote an inclusive, sustainable and well-being environment for citizens.

In worldwide cities such as Mexico City, the overcrowding of people in well-defined areas causes a series of health problems. Air pollution is one of the causes, so governments finance fixed air quality measurement stations in strategic points of the city. However, they are not enough for such complex cities. The research aims to develop an affordable and valuable system for society. The design takes the best of several Internet of Things technologies and systems, resolves incompatibility, minimizes costs, and tests with two public interest situations. The results show the behavior in days of environmental contingency and its uninterrupted operation for a reduced cost of USD$ 250 that can be visited on the web.

In this work we consider the effect of quantum entanglement regarding the reduction of energy in a wireless sensor network (WSN). Such theoretical networks are intended to use the phenomenon of quantum entanglement to reduce the overall energy consumption. As such the study allows to estimate the effective energy reduction and to compare with the energy consumption by a classical WSN.

Parallel flat plate electrochemical reactors are versatile devices that are used in a wide range of applications, including hydrogen production, organic compound synthesis, chlorine generation, wastewater treatment, and metal recovery. However, the flow dynamics within reactors are complex and can be difficult to measure. This study presents a novel analysis method for Parallel flat plate electrochemical reactors based on artificial vision techniques. The method uses a camera to capture images of the flow within the reactor, which are then processed using computer vision algorithms to extract quantitative information about the flow field. To validate the proposed method, the flow within the reactor was analyzed under three distinct configurations: empty channel, bifurcated, and canalized. The experimental results were compared with computational fluid dynamics (CFD) simulations using the mean squared error (MSE) metric. The experimental findings demonstrate the feasibility of the proposed approach.

One particular problem with ambulance location is when, after a disaster, there are no fixed bases for all the available ambulances, and the entity in charge must assign them a temporal location to efficiently provide emergency medical care. Most of the ambulance location problems are modeled as extensions of the set covering problem or the location problem. However, there is no efficient algorithm that can solve them in polynomial time, given that their solutions cannot be applied in emergency environments since they require a quick response. This paper describes a proposal to allocate temporary ambulance bases on streets that meet specific requirements. The contribution of this work is a methodology to reduce the search space of possible locations, simplifying the graph as much as it cans, and thereby will reduce the execution time when an algorithm to determine the locations is applied. The proposal includes a mathematical model integrated with two objective functions; one to minimize the distance between bases and demand points and the other to minimize the number of ambulance bases. This article presents the results of the methodology to reduce the search space, applied in a real-based study case, from 2017 when the strongest earthquake in Mexico’s history, collapsed several buildings in Mexico City. The emergency demand locations are carried out using Geospatial Information Systems (GIS) from an open-source platform. The methodology uses data prepossessing and an optimization algorithm to obtain the best available locations to reach the emergency demand points.

Cellular wireless networks have taken a preponderant role in modern society. With the emergence of 5G and 6G connections, the potential that they may unleash could transform the face in which mankind and machines work together. However, current 5G links are still scarce compared with the total amount of cellular users worldwide, and 6G is still in development phase. In this sense, 2G–4G links still dominate the market, with large physical infrastructures bearing transmissions ranging from 800 to 2,000 MHz. Thus, it is still important to provide reliable link budgets within such a frequency range in order to guarantee stability and quality of service. Despite there are many software-based calculators that provide a tool for link budgeting of cellular connections, they may be cumbersome to use, they could be of payment, they do not necessarily pose the used models as well as their range of validity, among other issues. The present work consists of the design and implementation of a calculation software tool for the construction of the link budget based on radio communications. The tool aims to offer ease of use, flexibility, accuracy, and accessibility in the area of communication systems, to obtain reliable and adequate link budget parameters, prior to the construction and commissioning of the real communications system. The software contains calculation options such as: conversion and display of basic measurement units for radio frequency links, Link Budget calculation, free space loss calculation applied to open environments, simulation and calculation of parameters for the design of communication systems, simulation of statistical models of wave propagation, among others. The software has a web-based friendly-user interface which can be used in any device and under any operating system, is modular and use generic processes, so it does not depend on specific transmission equipment.

The objective of this work is to create a system that allows a group of researchers to manage scientific information derived from their research activities in a safe way to promote interdisciplinary collaboration. Design Thinking (DT) was used for the development of the system. DT is focused on the continuous interaction of end-users with the development team. Each design step is centered on understanding user needs and motivations. The system implementation was done under the Scrumban methodology, using edge computing, microservices in the cloud, and artificial intelligence elements for information search. It was subdivided into two stages of development, with end-user feedback in each stage. The validation of the system was carried out with 16 researchers participating in a multidisciplinary project. The system lets each researcher control the information they share with their work leader, which, under a modular scheme, gives access to the publications to the other module members or modules. Users publish and consult shared documents and databases, and the system allows them to view, filter, calculate statistics, graph, and download the information. The system is available as a web application for this research group. However, it can be replicated to be used by other collaborative groups that seek levels of controlled access and portability of information within a secure environment, that is not constrained to a particular choice of hardware or operating system.

This paper describes an application of Wireless Sensor Network and Associative Models to monitor and forecast air quality in Smart Cities. The modifications that were made to the Gamma Classifier provide the foundation for this proposal. The improved model proposes a different way to measure similarity between patterns in the training set, reduces pattern encoding complexity, and improves forecasting performance on atmospheric data series. Experimental results and comparisons against other time series forecasting algorithms show that the proposed associative algorithm achieves better performance and makes better air quality predictions in urban settings.

Technology is a tool that has made it possible to make the world smaller, localized, and globalized, consequently, it has increased its use and appropriation in the current liquid society. In the educational context, in an educational institution in Colombia (level 9 of secondary), it is evident that the teachers of the institution are not trained to educate patients diagnosed with ADHD; consequently, this population is forgotten and treated like all students, therefore, this population is forgotten and considered outside the range of learning and causing damage on an emotional level.The observational study is carried out through subjective measurements and direct measurements with prior informed consent. The data analysis is carried out, by recording mixed data (qualitative and quantitative) that will allow us to know the behavior in real time. The data will be analyzed from contingency tables when they are a categorical variable, using chi-square tests (parametric variables); the analysis is performed through descriptive statistics, and inferential analysis by ANOVA or T- Student test. Consequently, UI (user interface) design faces a new challenge, in which UX (user experience) in adolescents is being used constantly. Achieving good design on digital platforms, web pages, and applications will have a positive impact on digital native adolescents; allowing them to improve their alternatives of communication and social interaction. This user-focused research for special populations answers the following research question: ¿How can the UI of ADH user search reduce execution time when using digital platforms, taking the educational environment as a case study?

The accurate classification of bees into pollen-carrying and pollen-free categories plays a crucial role in various aspects of bee research and management, avoiding economic losses in the bee-keeping sector and a reduction in pollinations of ecosystems. Although beekeepers can identify when pollen-free bees enter other hives and initiate possible looting, the task involves a lot of resources. In this paper, we propose a method for classifying images of bees based on their pollen-carrying status. We present two approaches: the first method utilizes a convolutional neural network (CNN) to classify original RGB images, while the second method enhances pollen regions in the images using digital image processing techniques before training the CNN. The results of the classification metrics demonstrate the effectiveness of both methods, with the second method achieving higher accuracy values and reduced loss compared to the first one. Moreover, the image enhancement techniques employed in the second method, including thresholding, morphological operations, and circularity ratio calculation, contribute to improved classification performance. Additionally, we discuss the CNN architecture, training parameters, and the significance of incorporating deep learning techniques in bee image analysis. The proposed method exhibits potential for application across different bee species, and future work may explore the extension of this approach to detect looting behavior.

We present and approach for monitoring and built a dataset of regional historical air quality data in Mexico City. We design a hybrid air quality network prototype that combines mobile and stationary sensors to collect street-level data on particulate matter (PM2.5 and PM10). The network is composed of mobile monitoring modules, both stationary at street level and mounted on vehicles, to capture a comprehensive sample of particulate matter behavior in specific areas. Collected data is transmitted using IoT network and processed using machine learning techniques, to generate predictive models to forecast air quality at street level. This approach is an additional improvement to current monitoring capabilities in Mexico City by providing granular street-level data. The system provides a regional and periodic perspective on air quality, enhancing the understanding of pollution levels and supporting informed decision-making to enhance public health and well-being. This research represents a solution for environmental monitoring in urban environments to know how the behavior from pollution levels in air is. The experiments show the effectiveness, and the model of forecast has an overall performance around 81% that is acceptable for the small geographical area testing. As future work is required to include a major number of nodes to collect data from a big geographical coverage and test with other models and algorithms.

Current binary and multi-class (family) approaches for malware classification can hardly be of use for the identification and analysis of other samples. Popular family classification methods lack any formal naming definitions and the ability to describe samples with single and multiple behaviors. However, alternatives such as manual and detailed analysis of malware samples are expensive both in time and computational resources. This generates the need to find an intermediate point, with which the labeling of samples can be speeded up, while at the same time, a better description of their behavior is obtained. In this paper, we propose a new automated malware sample labeling scheme. Said scheme assigns a set of labels to each sample, based on the mapping of keywords found in file, behavior, and analysis reports provided by VirusTotal, to a proposed multi-label behavior-focused taxonomy; as well as measuring similarity between samples using multiple fuzzy hashing functions.

The wireless communication mobile systems demand for compact and fully integrated radio frequency (RF) devices, low cost, small dimensions due to the space and volume available within the radio device is limited, and a high degree of miniaturization, capable to operate within the crowded 5G NR sub-6-GHz bands. In this way, the microstrip yagi patch antenna design, simulation, implementation, and measurement to be applied to 3.5 GHz band 5G technology is described and purpose of this work.

3D point clouds find widespread use in various areas of computing research, such as 3D reconstruction, point cloud segmentation, navigation, and assisted driving, to name a few examples. A point cloud is a collection of coordinates that represent the shape or surface of an object or scene. One way to generate these point clouds is by using RGB-D cameras. However, one major issue when using point clouds is the presence of noise and outliers caused by various factors, such as environmental conditions, object reflectivity, and sensor limitations. Classification and segmentation tasks can become complex when point clouds contain noise and outliers. This paper proposes a method to reduce outliers and noise in 3D point clouds. Our proposal builds on a deep learning architecture called PointCleanNet, which we modified by adding extra convolutional layers to extract feature maps that help classify point cloud outliers. We demonstrate the effectiveness of our proposed method in improving outlier classification and noise reduction in non-dense point clouds. We achieved this by including a low-density point cloud dataset in the training stage, which helped our method classify outliers more efficiently than PointCleanNet and Luo, S, et al.

The development of wireless communication systems has had a great increase due to the needs of digital services that are increasingly demanded by today’s societies. Wireless networks have become the most used systems by an increasingly growing number of users. In this context, the same wireless networks have diversified; proof of this are the solutions known as IoT (Internet of Things) based on LPWAN networks (Low Power Wide Area Network), which are adaptable, of low cost and relatively low development complexity, which have the potential to be used in a wide type of applications. Despite the advantages that they represent and the benefits that they could provide, there are still challenges to overcome in terms of design and implementation. One of the main aspects that require attention is the development and adaptation of an information (data) encapsulation structure according to the needs of the system and the application, since such an information must be transported safely, fully and efficiently, always respecting the limitations of the communications system. This paper proposes a data framing protocol structure based on an LPWAN/IoT monitoring system, which is based on COTS components and has the function of acquiring environmental variables.

Wireless sensor networks are used extensively to monitor different environments and physical variables. However, in many cases, there are many nodes in the same region and given its distributed nature, these networks can suffer from corrupted data if too many transmissions occur at once. Therefore, it is important to maximize the number of successful transmissions while also minimizing the amount of energy used to transmit this data. The main component of wireless sensor networks that impacts these variables is the transmission probability. Thus, it follows that to improve successful transmission probability and to reduce energy consumption, an adequate transmission probability value should be selected. In particular, we propose the use of a transmission probability that decrease as the energy consumption increases in such a way as to reduce energy consumption towards the end of the system lifetime but that still allow the reporting of the events, i.e., packet transmissions.To this end we propose and evaluate the system performance in terms of average energy consumption and successful packet transmission probability using different mathematical functions for the transmission probability that decrease as the energy consumption increases but do not decrease to zero, allowing sporadic transmissions towards the end of the system operation and compare them to the case of using a fix value.

Sentiment analysis aims to extract general information from texts and understand the opinion, attitude, or emotion of an individual or group of people towards a specific topic. Currently, a source of information used is the messages published on Twitter, which offer possibilities of great interest to evaluate the currents of opinion disseminated through this social network.However, the enormous volumes of text require tools capable of automatically processing these messages without losing reliability. This article describes a technique to address this problem. This technique uses a variant of a Bi-GRU, a recurrent neural network (RNN) that promises to highlight local and global contextual features of tweets to increase the accuracy of classifying opinions in that social network. Experiments show better performance in tweet analysis, improving accuracy, recall, f-score, and accuracy parameters than traditional techniques. Finally, we identify the advantages and limitations of the system for its application to research on “Pueblos Mágicos” tourism in Quintana Roo.

This work compares popular classifiers based on pattern recognition techniques of supervised learning, including k-Nearest Neighbors, Naïve Bayes, Support Vector Machines, Artificial Neural Networks, and Decision Trees. Such techniques are applied to a dataset related to vertebral column orthopedic diseases. Different parameter values employed by each classifier are tested, resulting in an accuracy of around 80 $$\%$$ % in almost all approaches, where the k-Nearest Neighbors alternatives were the most accurate. Finally, a brief discussion of particular highlights of how the metrics affect the performances of the classifiers is presented.

There is no doubt that nowadays, the use of smartphones for communication between two or more entities through instant messaging applications has become a trend model in our society. New messaging applications started to emerge and try to replace traditional SMS. In fact, they have become the main communication route, and it is almost impossible to find someone who does not use at least one of these kinds of messaging applications. However, building them with security and privacy in mind of developers was not important in the beginning. In other words, when the popular messaging applications were created, they did not support end-to-end encryption, only standard client-to-server encryption, which gave the service providers access to more private information than necessary. Additionally, information that is exchanged in such instant messaging applications has the characteristic to be sensible that results in the necessity to achieve security services associated with users information, by achieving confidentiality, integrity, and authenticity in sent and received messages. In this article, we present a security verification on the Signal and MTProto 2.0 cryptographic protocols, which are contained in the most commonly used instant messaging applications. The security verification is made by using automatic verification tools and obtained results show that the protocols are flawless in terms of their construction, message delivery logic, and semantics. In other words, they are safe from attacks that automatic protocol verification tools check for.

During the last few years, catastrophic accidents occurred at Petroleos Mexicanos (Pemex), which led to the company working on implementing A Safety, Health, and Environmental Protection Program that contribute to reducing them. To be able to provide the hydrocarbon, oil, and petrochemical pipeline transportation service improvements, the 12 Best International Practices (MPI) of the SSPA System (Safety, Occupational Health, and Environmental Protection) are used to guarantee the safety of the personnel that their works. The present work proposes to develop an information system that allows monitoring one of these 12 MPI called operational discipline, to guarantee rigorous monitoring and compliance of all procedures and operational, administrative, and maintenance instructions of the Work Centers through a methodology that comprises five phases for the development of this information system to Transportation, Maintenance, and Pipeline Services Management operational discipline by collecting the information in a system developed exclusively for this purpose that documents the cases of incidents present in the areas, which records, monitors and controls that documents the operational discipline process shows containing and evaluating the 5 stages of number 6 Best International Practices to standardize the procedures reported by each area.

Sensitive personal information is at risk of exposure by the institutions it is shared. Institutions are responsible for preserving the privacy of the personal data they hold, even more so, in the case of sensitive data. This paper shows the design of ICIS, a model that considers the context to identify 55 personal data types in unstructured texts of government type documents, regardless the size and type, and then classify each text segment as sensitive personal information, using natural language processing and machine learning techniques. ICIS not only indicates whether a text segment contains sensitive information or not, it also indicates personal data identified in each text segment, their location in the document and whether each text segment is classified as sensitive information. The main contributions of this work are both the identification of personal data and the classification of sensitive information based on the context, and the definition of sensitive personal information, in computational terms.

Currently, education presents important challenges due to the Pandemic we are going through due to SARS-COV-2, which has generated uncertainty in all aspects; hence the importance of incorporating leaders who provide confidence and guidance to their colleagues within organizations has become an indispensable factor. Therefore, this article analyzes leadership styles (positive, ethical, authentic, spiritual, service, and transformational Leadership) as case studies observed in Universities during the Pandemic to identify the leadership style that helps reduce uncertainty in this Pandemic.

Adolescents and young adults are increasingly acknowledged as mental health vulnerable populations whose mental health is particularly vulnerable. Conditions such as anxiety and depression are especially prevalent. The detection of these conditions and timely support in educational institutions remains a challenge, as they can have an important impact on the well-being of students and academic performance. In this article we present a new microblogging web platform designed to solve these problems by providing students with a safe, expressive space to freely communicate about topics of interest in their respective communities, providing the chance to do so. This platform encourages users to share their experiences, opinions and concerns. One of the main features of this platform is its seamless integration with various modules that capture valuable metrics of mental health. By linking the results of text analysis with data from commonly used questionnaires for mental health symptom detection, a comprehensive picture of an individual’s well-being can be established. In particular, we developed a support vector machine model specifically trained to identify depressive symptoms in the written language shared on the platform. Our findings demonstrate the effectiveness of the microblogging platform in conjunction with the classification model. By leveraging this approach, educational institutions can gain insights into students’ mental health by analyzing their written expressions. This integrated system offers a promising strategy for detecting and addressing depressive symptoms among adolescents and young adults, ultimately contributing to improved well-being, academic performance, and targeted support in educational settings.

This study presents the implementation of a feed-forward neural network (FNN) for the classification of household appliances, specifically refrigerators and microwave ovens. The data used in this study was collected using a PZEM-004t sensor, capturing various parameters such as current, power, power factor, frequency, voltage, and energy consumption. The collected data was then preprocessed, including outlier removal, handling missing data, and applying normalization techniques. The FNN model was trained using labeled and sorted data, with current, power, and power factor chosen as input parameters due to their high variation and relevance. The neural network was trained using MATLAB’s neural network toolbox, employing a configuration with 12 input nodes, two hidden layers, and an output layer. The model achieved impressive performance during the training stage, with accuracy, fallout, recall, miss, and specificity metrics calculated. In the generalization stage, the model achieved an accuracy of 94% and demonstrated a correct identification rate of approximately 95% with a false positive rate of around 5%. The Area Under the Curve (AUC) value, calculated as 0.9692, further validated the model’s robustness and accuracy in appliance identification tasks. The study concludes that the trained neural network has great potential for accurate appliance identification in residential settings, laying the foundation for intelligent energy management practices.

Learning environments and interactions in educational contexts generate transformations in student learning outcomes, in which conceptual, methodological, and investigative trends are identified; from those assumed by learning environments and their mediations, conceptions, origins, and characteristics that emphasize those who learn, knowledge, and the educational community. Based on this research work, an app with augmented reality about autopsies is proposed, used by professionals in health sciences (semester 11) when considered for learning. This look proposes a space constituted by two design experiments and they were correlated with the conceptual design of an interactive digital learning ecosystem, in this way the learning environments as an object of the study propose a domain articulation space as a theoretical and practical perspective about the complexity of educational technological phenomena; operationalized in different phases: planning, learning outcome mediated by (personal learning environmental (AVA virtual learning environment – didactic sequence), and reflection on the teaching pedagogical praxis. At the same time, digital technologies are in the process of expansion and generalization in educational systems, allowing communication between students and teachers. Therefore, it is necessary to think about new pedagogical practices that promote knowledge processes, dynamic practices thought from the encounter in otherness and interactions with a multi-situated thought or group cognition, to the bidirectional relationships between systems, virtual and human educational environments.

The cybersecurity risk audit is a relatively new field. The objective of cybersecurity risk audits is to identify deficiencies or deviations in cybersecurity countermeasures. Currently, cybersecurity risk audit guidelines do not include an internal control approach that aligns with cybersecurity standards such as ISO 27001 or the NIST CSF. Internal control is essential for addressing cybersecurity risk audits. This article proposes a cybersecurity risk audit guideline called CRAG (Cybersecurity Risk Audit Guideline), created using the SADT (Structured Analysis and Design Technique) model. CRAG aims to be comprehensive in the various applications that a cybersecurity risk audit guideline can have. The CRAG guideline consists of seven steps and 28 activities, as well as the content that the resulting audit report should include. Additionally, this article provides guidelines for its proper implementation, as well as examples of its potential applications.

Adversarial attacks are designed to disrupt the information supplied to Artificial Intelligence (AI) models causing a failure in their intended purpose. Despite late state-of-art performance on computer vision tasks like image classification, AI models can be vulnerable to adversarial attacks. Hence, to mitigate this risk the AI models require additional steps. This research presents the implementation of a novel bio-inspired defense method against adversarial attacks. The defense is based on the use of a deterministic monogenic layer at the top of a ResNet-50 convolutional neural network model used for image classification with the CIFAR-10 data set. The results show the ConvNet with the bio-inspired defense approach outpacing the regular convolutional architectures, this can be translated into increased resistance to adversarial attacks and greater reliability of artificial intelligence models. The Structural Similarity Index Measure (SSIM) and the Peak Signal-to-Noise Ratio (PSNR) metrics were measured on the bio-inspired layer activations to obtain a quantitative explanation of the improvement. The results obtained from this research expose the potential performance of the monogenic layer in confronting adversarial attacks and encourage further expansion of the knowledge in the field of the monogenic signal for AI purposes.

Inventory control is an element of great importance for the development of large, medium and small companies, for this reason the main objective of this study is to create an app implementing QR code for inventory control in Air Force warehouses. Colombiana, which will allow carrying out the inventory management method in the warehouses of the Colombian Air Force, making use of technological tools in the globalized world to facilitate control in this type of areas and functions. This will allow for a more rigorous control of inventory administration and management of administrative processes for the warehouses of the aeronautical logistics center, miscellaneous, aerial weapons and ground weapons, thereby generating doctrine in the logistics process, standardizing activities and controls for the inventory management and warehouse control. This leads to obtaining as a result a correct decision making facilitating the execution of the activities of the warehouses that will allow having an essential strategic stock to guarantee the operations of the Colombian Air Force, in the same way it reflects the opportunity to improve the operation of this area of great importance for the daily operation of the institution’s activities. This being essential for the correct development of the institutional mission, ensuring that we always have the products regardless of the conditions that may arise, such as a possible war.