2026-02-15
Apolinar Velarde Martínez
Pork consumption has increased considerably in recent years, making pig reproduction of vital importance. Feeding sows during the gestational period involves several factors that impact their health, well-being, and postpartum outcomes. Automated feeding systems for sows, unlike feed delivery by human feeders, can measure and adjust the quantity and quality of feed provided at scheduled times with diets established by human experts, allowing for the establishment of a nutritional program for each stage of pregnancy. They also allow for the calculation, using probabilistic and statistical tools, of future diets for the gestational period, which promotes the health and well-being of the sow. This research paper describes, and presents the results obtained from experiments carried out with intelligent automatic feeders for pregnant sows, implemented with Internet of Things (IoT) devices and operated with a predictive computing algorithm, which uses decision trees to predict feeding for the gestation stages of sows, using the weight of the sows, as well as the amount and type of feed provided to the sow. The experiments were carried out with a typical feeding process performed by a human operator, against a feeding process provided with automatic feeders to two different sets of pregnant sows. The results obtained with the automatic feeders show weights closer to the standard weights established for the gestational stages of the sows. Smart citations: https://scite.ai/reports/10.61467/2007.1558.2026.v17i2.1260 Dimensions . Open Alex .
2026-02-15
Luis Alejandro Reynoso-Guajardo, Axel Roberto-Perez, Carlos Hernandez-Santos, Amadeo Hernandez, José Isidro Hernández-Vega, Mario Carlos Gallardo-Morales, Nain de la Cruz, María Ernestina Macias-Arias, Jorge Eduardo Ortega-Lopez, Juan Velazquez-Coronel
This study introduces a predictive air quality monitoring system based on Random Forest machine learning models and low-cost embedded sensors. The system was designed and implemented in Guadalupe, Nuevo León, Mexico, to monitor carbon monoxide (CO), car-bon dioxide (CO2), and particulate matter (PM). Real-time data was collected using a Particle Photon 2 microcontroller with four different sensors. The data was processed using Python scripts, and the Random Forest model was trained to predict future pollutant values. Results demonstrated strong model performance, validated through statistical evaluation metrics and graphical comparisons. The proposed system shows promise for deployment in smart urban environments. Smart citations: https://scite.ai/reports/10.61467/2007.1558.2026.v17i2.1234 Dimensions . Open Alex .
2026-02-15
Rafael Castaneda-Diaz, Daniela López-Betancur, Carlos Guerrero-Méndez, Efrén González-Ramírez, Salvador Gómez-Jiménez, Flossi Puma-Ttito
This paper presents an analysis of the quality and randomness of information in images generated with a convolutional autoencoder (CAE). The CAE convolved altered color images from CIFAR-10 dataset. The CIFAR-10 images were altered by randomly setting 30%, 60%, and 90% of pixel values to [0, 0, 0] or [255, 255, 255], respectively. In the validation stage, Mean Square Error (MSE) loss function reached 0.0115 and the accuracy metric 0.7461. Similarity Structural Index Measure (SSIM), Pearson Correlation Coefficient (PCC) and Peak Signal-to-Noise Ratio (PSNR) metrics, assessed quality of generated images. The assessment results ranged as follows: SSIM [0.3251, 0.6830], PCC [0.5034, 0.9358], and PSNR [18.63 dB, 26.04 dB]. The Shannon metric assessed randomness both locally and globally for each image, ranging from 1.25 to 2.49 bits, and from 6.61 to 9.71 bits, respectively. CAE implementations highlight their potential for applications in technological innovation. Smart citations: https://scite.ai/reports/10.61467/2007.1558.2026.v17i2.952 Dimensions . Open Alex .
2026-02-15
Luis Alberto Hernández Montiel, Edmundo Bonilla Huerta, Roberto Morales Caporal, Carlos Bueno Avendaño
In this paper, we present a new algorithm that combines Hu invariant moments, data preprocessing techniques, and fuzzy logic to analyze and classify signals obtained from the macro and micro head movements that a person makes. Firstly, we project different videos to elicit the disgust emotion in various participants. Following the stimulation, we recorded all the participants' reactions on video. The next step is to apply Hu moments to transform the different movements that the participant made to create a signal. This process yields polynomial regression and applies normalization methods to standardize the size and numerical scale of all signals. We implement a fuzzy inference system to classify the signals. The results presented by this algorithm show high performance in sorting movements to detect disgust emotion. Smart citations: https://scite.ai/reports/10.61467/2007.1558.2026.v17i2.1120 Dimensions . Open Alex .
2026-02-15
Ángel J. Sánchez García, Álvaro Barradas Fernández, Oscar Alonso Ramírez, Xavier Limón
Cluster analysis is an unsupervised machine learning approach that groups data into homogeneous categories without the need for predefined labels. Although it was not originally developed for software engineering, this technique has increasingly been applied to support various activities in the software design phase. However, information about its use remains scattered across different studies. To address this gap, this work presents a systematic literature review synthesizing the state of the art on the application of cluster analysis in software design. Following a rigorous selection process, 14 primary studies published between 2019 and 2025 were identified from four digital libraries: IEEE Xplore, ACM Digital Library, Springer Link, and ScienceDirect. This review highlights the contexts in which clustering has been applied, emphasizing its predominant role in class decomposition tasks and the frequent adoption of the K-means algorithm, while also documenting the algorithms and tools used during design activities. Furthermore, the analysis discusses the benefits and challenges of adopting cluster analysis in this stage of development. The findings provide software engineering researchers and practitioners with a consolidated overview of the role of cluster analysis in software design, offering insights into its potential, limitations, and directions for future research and practice. Smart citations: https://scite.ai/reports/10.61467/2007.1558.2026.v17i2.1262 Dimensions . Open Alex .
2026-02-15
Carlos Enrique Greene-Mex, Antonio Armando Aguileta-Güemez, Jorge Alberto Ríos-Martínez, Raúl Antonio Aguilar Vera
Understanding dogs’ emotional patterns through their tail movement is key to strengthening the human-canine bond and improving their well-being. This study presents an innovative approach to identify the direction of tail movement (either left or right), using the dog’s hip as the primary reference point. Through a preprocessing and feature extraction process, a Support Vector Machine (SVM) classifier was trained with spatial data from the hip and compared to a classifier from a previous study. The results indicate that the classifier trained with hip data achieved a 99% score in accuracy, precision, and F1-score metrics. Additionally, the Friedman test was performed to verify whether there are statistically significant differences between the two classifiers. Smart citations: https://scite.ai/reports/10.61467/2007.1558.2026.v17i2.1252 Dimensions . Open Alex .
2026-02-15
Gilberto Bojorquez Delgado, Jesus Bojorquez-Delgado, Manuel A. Flores-Rosales
Accurate estimation of the Normalized Difference Vegetation Index (NDVI) is crucial for precision agriculture and environmental monitoring. This study compares five machine learning algorithms LSTM, CNN-LSTM, XGBoost, Random Forest, and Gradient Boosting to predict NDVI using time series of multispectral Sentinel-2 data in a maize crop in Guasave, Sinaloa, Mexico. After data preprocessing, the models were evaluated using cross-validation and metrics such as MSE, MAE, RMSE, MAPE, and R2. The results showed that the LSTM model achieved the best performance in accuracy, while tree-based models struggled to predict extreme values. Recurrent neural networks demonstrated a greater capacity to capture complex temporal dependencies, although they still require improvements to optimize their robustness and precision. Smart citations: https://scite.ai/reports/10.61467/2007.1558.2026.v17i2.597 Dimensions . Open Alex .
2026-02-15
José Félix Serrano Talamantes, Mauricio Olguín Carbajal, Gerardo Miramontes de León, Héctor Durán Muñoz, Claudia Sifuentes Gallardo, Carlos Avilés Cruz, Gabriel de Jesús Celis Escudero
This proposal shows a methodology designed for the process of detecting emotions through facial features. The process of facial features detection involves several stages among which the most important ones are: Acquisition of the training set of images, detection and segmentation of the face using face location techniques in images using Viola & Jones algorithm. We also make use of neural networks that are part of Deep Learning techniques. In this way we propose to recognize people’s faces and also perceive their emotions through gestures that are captured by means of a camera, allowing us to obtain these in soft real time. The processes of this methodology were developed and programmed in the MATLAB and Python code. There was a significant improvement in the recognition results throughout the CNN. Smart citations: https://scite.ai/reports/10.61467/2007.1558.2026.v17i2.1261 Dimensions . Open Alex .
2026-02-15
Alejandro Fuentes-Penna, Ricardo A. Barrera Cámara, Raúl Gómez Cárdenas, Óscar Daniel Hernández González, Aristeo Castro Rascón
Based on advances in computing and devices as an extension of the human body, new technologies focused on neuromodulation and with the help of artificial intelligence, can generate lucid dreams and even improve concentration. With noise cancellation through devices such as smart hearing aids or with the generation of devices that detect and modify brain waves, the human body has been empowered towards mental conditions and, in a sense, towards improving the quality of life through sleep. of life through sleep. However, we would be at a point where the ethics and philosophy of technology ethics and the philosophy of technology converge to identify the use of such devices to improve human devices to improve human conditions without overstepping the legal and ethical realm in their use. Smart citations: https://scite.ai/reports/10.61467/2007.1558.2026.v17i2.1135 Dimensions . Open Alex .
2026-02-15
Viridiana Barrera Soto, Juan Carlos Huerta Mendoza, José Lázaro Martínez
Diabetes is a chronic metabolic disease characterized by elevated levels of glucose in the blood (or blood sugar), which over time leads to severe damage to the heart, blood vessels, eyes, kidneys, and nerves. The most common type is type 2 diabetes, usually in adults, which occurs when the body becomes resistant to insulin or does not produce enough insulin. By using artificial intelligence (AI) techniques in complex problems such as disease diagnosis, a degree of certainty in the results has been achieved to identify a specific type of disease. These applications have been advantageous because large amounts of patient data can be analyzed to find patterns. This work proposes a platform for the prediction of type 2 diabetes based on clinical or personal indicators. To do this, two supervised classification models were constructed using the PIMA Indian Diabetes dataset and the Centers for Disease Control and Prevention (CDC) dataset, integrating both into a web platform for prediction with new data to support the decisions of doctors and healthcare professionals. By integrating different algorithms into the final predictive model through voting weighting, the accuracy percentage in prediction has been increased. Smart citations: https://scite.ai/reports/10.61467/2007.1558.2026.v17i2.1190 Dimensions . Open Alex .
2026-02-15
Jorge Ramos Martinez, Juan Carlos Huerta Mendoza, David Tomas Vargas Requena, Ramón Gerardo Tijerina Rodríguez, Sanjuanita Crescencia Ortiz Valadez
This project addresses a significant challenge in the formation of sports teams within the university environment of the Reynosa Rodhe Multidisciplinary Academic Unit (UAMRR) of the Autonomous University of Tamaulipas (UAT). The project identifies the need to establish balanced and competitive teams that reflect the diversity of skills in sports activities. The project proposes the development of a genetic algorithm aimed at the automatic generation of optimal sports teams using techniques based on natural evolutionary processes, such as selection, crossing, and mutation. This algorithm is integrated into a web application with a database that stores detailed information about the students. It is developed in an intuitive design to ensure an optimal user experience and facilitate application navigation. Smart citations: https://scite.ai/reports/10.61467/2007.1558.2026.v17i2.1192 Dimensions . Open Alex .
2026-02-15
Kevin Jorge Montes Lorenzo, Francisco José Moo Mena, Antonio Armando Aguileta Güemez
In a previous study, we compared the performance of two well-known models, YOLOv8 and RT-DETR, for firearm detection. The results showed that YOLOv8 achieved superior performance, while RT-DETR also produced significant results. These findings suggested the potential to further improve detection by exploring result fusion methods. Unlike the original comparative analysis, this work investigates how integrating the outputs of multiple detectors can enhance both accuracy and robustness in firearm identification. The existing literature on result fusion, as explicitly applied to this field is scarce, leaving a promising line of research open. In this context, strategies such as averaging results, selecting the best detector, and time-weighted as well as real-time weighted methods are analyzed. The objective is to demonstrate that result fusion represents an effective method for enhancing the performance of object detection systems in complex scenarios. Smart citations: https://scite.ai/reports/10.61467/2007.1558.2026.v17i2.1269 Dimensions . Open Alex .
2026-02-15
Cecilia González Servín, Christian E. Maldonado Sifuentes, Olga Kolesnicova, Grigori Sidorov
This work evaluates how domain adaptation affects Transformer-based neural machine translation (NMT) for the low-resource Purépecha–Spanish pair. Building on a system fine-tuned on a verse-aligned Bible corpus, we introduce an out-of-domain grammar-book dataset (1,626 sentence pairs: 1,297 used for adaptation, 329 held out for testing) to quantify (A) zero-shot transfer (Bible→G-test) versus (B) adaptation (Bible+G-train→G-test). Using BLEU and ROUGE, zero-shot performance is weak for Marian (BLEU=0.2272) and mBART-50 (BLEU=1.9992), revealing substantial domain mismatch. After adaptation, scores rise sharply: Marian reaches BLEU=21.2699, mBART-50 achieves BLEU=28.8776, with parallel gains in ROUGE (e.g., mBART-50 ROUGE-L=0.5791). Qualitatively, adaptation reduces repetitive/degenerate outputs and improves handling of metalinguistic terminology and everyday constructions. These results show that multilingual pretrained Transformers + lightweight in-domain data provide strong improvements for low-resource NMT under domain shift and highlight the value of diverse domains and speaker-informed evaluation. Smart citations: https://scite.ai/reports/10.61467/2007.1558.2026.v17i2.1265 Dimensions . Open Alex .
2026-02-15
Fevrier Valdez, Hector M. Guajardo, Oscar Castillo, Patricia Melin, Prometeo Cortes-Antonio
This paper explores a hybrid approach combining the Dragonfly Algorithm (DA) and Firefly Algorithm (FA) to balance exploration and exploitation, avoiding local optima and refining solutions in promising areas. The hybrid achieved higher-quality solutions, faster convergence to optimal results, and adaptability to diverse optimization problems through complementary strategies. Additionally, the Cuckoo Search Algorithm (CS), known for its effectiveness in global optimization via random search and solution space exploitation, was integrated. To further enhance performance, Type-2 Fuzzy Logic was applied for parameter adaptation in the algorithms. Smart citations: https://scite.ai/reports/10.61467/2007.1558.2026.v17i2.872 Dimensions . Open Alex .
2026-02-15
Paola Itzel Delena-García, Yenny Villuendas-Rey, León S. Mora-Guerrero, Antonio Alarcón-Paredes
Biomedical datasets often contain noise, missing values, imbalance, and heterogeneous feature structures, making them difficult to model reliably and complicating the extraction of discriminative patterns required for effective classification. Although modern machine-learning models can achieve strong performance on such data, many of these approaches operate as opaque systems, offering little insight into how decisions are produced—an essential requirement in biomedical applications. This work introduces the Explainable Artificial Immune System (XAIS), an immune-inspired classification model that delivers prototype-based explanations derived from similarity-driven antibody responses and complemented by performance-aware indicators, providing users with direct evidential insight into each decision. XAIS was evaluated on eight publicly available biomedical datasets using stratified 5-fold cross-validation and compared against standard machine-learning classifiers. The results show that XAIS attains competitive predictive performance while offering structured, instance-level evidential explanations, underscoring its potential as a transparent and trustworthy foundation for biomedical decision-support systems. Smart citations: https://scite.ai/reports/10.61467/2007.1558.2026.v17i2.1278 Dimensions . Open Alex .
2026-02-15
Adán Jiménez-Montoya, Juan Benito Pascual-Francisco, Orlando Susarrey-Huerta
This paper presents an innovative method for qualitative reasoning in the conceptual synthesis of mechanisms, leveraging AI-derived knowledge-based principles. The methodology allows for discretizing a mechanism's overall behavior without computational implementation. Relative motion between components, represented as qualitative states, is captured in a qualitative motion vector. These vectors form a general movement matrix that characterizes the mechanism's behavior, providing insights into component functions, movements, and transitions. By using ratchets as restriction functions, underlying behaviors are isolated from the matrix. These behaviors are used to generate conceptual designs for new mechanisms fulfilling specific kinematic functions. A case study is presented: synthesizing a mechanism that converts oscillatory rotation into unidirectional rotation using a differential gear train. The qualitative behavior of the resulting design is visualized in a vector diagram and compared with a CAD simulation. This method provides a knowledge base for training AI models in conceptual synthesis without needing specialized software. Smart citations: https://scite.ai/reports/10.61467/2007.1558.2026.v17i2.1208 Dimensions . Open Alex .
2026-02-15
Abraham Jorge Jiménez Alfaro, Norma Karen Valencia Vázquez, Griselda Cortés Barrera, Edgar Corona Organiche
The ISO/IEC 90003:2014 standard oriented to the application of quality management in software development processes, requires a structured evaluation that considers technical and organizational criteria. Its purpose is to support organizations that develop or maintain software to implement an effective quality management system using the structure and requirements of ISO 9001:2008, specifically adapted to software-related activities. This article proposes a hybrid heuristic computational model based on the PMP-Greedy (Positional Major Mean Weight) and AHP (Analytic Hierarchy Process) methods to assign weights to the categories that make up a framework aligned to ISO/IEC 90003:2014 that allows companies engaged in software development to measure the degree of compliance with the standard, as well as, to assess conformance and non-conformance to achieve certification to the standard. Smart citations: https://scite.ai/reports/10.61467/2007.1558.2026.v17i2.1155 Dimensions . Open Alex .
2026-02-15
Joel Urquiza-Martínez, Gabriel González-Serna, Andrea Magadan-Salazar, Nimrod González-Franco, Jonathan Villanueva-Tavira
Webcam-based gaze tracking has emerged as a cost-effective alternative to infrared eye-tracking systems; however, its robustness under real-world conditions remains limited, especially when monocular (single-eye) models are employed. This study offers a systematic comparison between single-eye (SET) and double-eye (DET) convolutional neural network architectures for appearance-based gaze estimation utilizing standard webcams. A lightweight CNN processes normalized eye-region images to predict gaze coordinates on the screen and is evaluated through two complementary protocols: (i) a static 9-point calibration and (ii) a dynamic trajectory-following task (“blue-ball”) that traverses screen edges and corners. To enhance generalization, the dataset incorporates controlled data augmentation—pose variation, illumination changes, blur/noise, and partial occlusions—and employs a participant-stratified experimental design. Robustness and temporal consistency are further improved by integrating a Kalman filter for trajectory smoothing and a DBSCAN-based clustering calibration stage that suppresses outliers and stabilizes gaze estimates. Performance metrics include pixel error, angular error, and trajectory stability. Under identical training and evaluation conditions, the DET model consistently surpasses the SET model, attaining lower spatial error and smoother trajectories, particularly under asymmetric lighting and partial occlusions. Utilizing only a consumer-grade webcam and a lightweight CNN, the proposed methodology delivers real-time performance without infrared sensors or proprietary licenses, thereby providing an accessible, reproducible, and robust solution for gaze estimation in human–computer interaction applications. Smart citations: https://scite.ai/reports/10.61467/2007.1558.2026.v17i2.1304 Dimensions . Open Alex .
2026-02-15
Rene Davila, Rocio Aldeco Perez, Everardo Barcenas
Smart Contracts are stored and executed on a Blockchain network, thereby automatically enforcing the predefined rules once the execution conditions are satisfied. Hence, if the contract incorporates contradictory design rules, it may result in unforeseen outcomes within the blockchain environment. Accordingly, this proposal models the rules embedded in a Smart Contract through the Web Ontology Language (OWL), by applying the formal definition of consistency within a verification framework grounded in Description Logics. Smart citations: https://scite.ai/reports/10.61467/2007.1558.2026.v17i2.1258 Dimensions . Open Alex .