. Colima’s volcano is one of the most active volcanoes in Mexico, therefore, constant monitoring of its explosive activity is necessary to provide a timely warning of volcanic danger for nearby populations. Each explosive event generates seismic waves. The characteristics of the seismic signals can serve to identify an explosive event as strong, or as a weak explosive event. Therefore, the rapid identi(cid:28)cation of the type of explosive event can be important for estimating the level of danger for nearby populations. Normally, the identi(cid:28)cation of the type of seismic signal produced by the explosions is carried out after the event has ended (o(cid:27)-line analysis). An improvement to the o(cid:27)-line type of analysis is to perform an analysis at the moment the continuous explosive seismic signal originates ( on-line analysis) since this will allow determining the type of seismic signal in one stage early where the event originates. An algorithm was developed that allows identifying the type of seismic signal through analysis (on-line) and issuing an alarm to the population of the occurrence of a strong explosive event with a high probability of danger. The results obtained by our on-line analysis algorithm show that the type of seismic event can be identi(cid:28)ed in the same way as using the o(cid:27)-line method.
Resumen. Atribución de Autoría es la tarea de identificar el autor de un texto anónimo. El uso de inteligencia artificial es muy común en esta tarea y uno de los principales problemas es que no se cuenta con suficientes textos de entrenamiento de uno o más candidatos autores, lo que genera un problema de desbalance, afectando en gran medida el rendimiento de los métodos propuestos en el estado del arte. La mayoría de estos métodos utilizan SVM como clasificador y una de las siguientes características como modelo de representación de texto: unigramas de palabra o trigramas de carácter. En este artículo se propone un método con un rendimiento superior a los métodos del estado del arte en muestras balanceadas y desbalanceadas, esto se logra mediante la combinación de distintos tamaños de n-gramas para formar la bolsa de n-gramas y generar un nuevo modelo de representación de texto. Los experimentos realizados muestran la importancia de entrenar al clasificador con una bolsa de n-gramas de diferentes tamaños y no solo con un tamaño fijo. Los mejores resultados se obtienen cuando se combinan unigramas y bigramas de palabra, además se muestra la importancia que tiene la estrategia de clasificación cuando se utiliza SVM. Palabras clave:
. Lane detection is an important module for active safety systems, as it increases safety and reduces traffic accidents caused by driver inattention. Illumination changes or occlusions make lane detection a challenging task, especially if the detection is performed from a single image. Consequently, this paper presents a probabilistic approach based on the Kalman filter, which using information from previous image frames manages to estimate the lane that could not be detected in the current image frame, considering uncertainty in the prediction as well as in the detection. To this end, a principal component analysis of the segmented curvature is introduced with the purposes of dimensionality reduction, moving from a large dimensional pixel representation to a considerably reduced space representation.
. The current context in which higher education institutions operating in Mexico is adverse and multifactorial. In this work, the data obtained from a survey applied to 1,582 students are analyzed to determine the main factors that influence school dropout in a pre-COVID19 stage. With this information, an analysis of the decision tree was developed, detecting the main routes that influence school dropout. This study can be useful both to the public and to the instances involved in decision-making, to try to create an environment conducive to allow students to continue with their university education.
Resumen. El modelado de procesos afectivos en sistemas artificiales contribu-ye a generar comportamientos m´as cre´ıbles en sistemas interactivos como los agentes conversacionales personificados. El presente art´ıculo describe un modelo computacional de regulaci´on de emociones basado en la teor´ıa de J.J. Gross la cual propone cinco estrategias de regulaci´on emocional. Adem´as, el modelo propuesto considera las diferencias individuales en la implementaci´on de cada estrategia de regulaci´on a partir de los cinco tipos de personalidad tomados de los cinco grandes rasgos (Big-Five). La implementaci´on de este modelo y su integraci´on con una arquitectura computacional de emociones existente se describe en detalle y se presentan los resultados iniciales a partir de un conjunto de simulaciones representando un escenario interactivo con agentes conversacio-nales personificados
. In the analog circuit design area, it is common to work on pre-established topology templates, in which the circuit elements and their
Resumen. League of Legends es un videojuego de estrategia competitivo 5 vs 5, se puede observar un amplio margen de personajes (más de 150) disponibles para escoger; existe un interés importante en conocer el resultado de una partida antes de comenzarla, y aunado a ese interés una creencia popular, “Las partidas se pueden ganar desde la selección de personajes, ya que existen tipos de personajes que son mejores contra otros”. Abriendo la posibilidad de poder predecir el resultado de una partida con solo saber sus condiciones iniciales, basándose en los personajes. Aquellas personas que pudieran acceder a esta información antes de comenzar a jugar tendrían una ventaja crucial. Para corroborar esta hipótesis se recurrió al uso de Deep Learning con el modelo de red neuronal “Perceptrón Multi Capa”. Mediante una base de datos con un total de 62 mil partidas, las cuales fueron obtenidas directamente de la API oficial de la desarrolladora del videojuego RIOT GAMES
. For the development of a credible virtual agent, it is important that this agent has characteristics to create a more immersive experience for users and increases the effectiveness of communication. One of these characteristics is the ability to interact through verbal and non-verbal communication. That is why in this work an integration model of dialogue, facial and gestural expressions are proposed, where it is essential to consider the attributes of emotions and personality because these directly influence the expressions. This paper proposes the use of an optimization method named ELECTRE III for the process of selection of dialogue, facial and gestural expressions, using a corpus characterized by diverse criteria based on the influence of personality and emotions.
. In Mexico, according to INEGI data, the main causes of death are Heart Disease (CD) and Diabetes Mellitus (DM). In the first semester of 2021, 579,586 deaths were registered, occupying the first places only after COVID-19.
. There are 1.45 billion vehicles in the world, while in Mexico there are 50 million. Most are family vehicles with up to 5 passengers. On the other hand, in Mexico there are 10 million children between 0 and 4 years old. In multiple events it has happened that children have been left inside the car. Infants have
. Steganography through deep learning techniques has been widely driven through Generative-Adversarial Neural Networks. This paper presents the implementation of the ISGAN model to embed (hide) a QR code in another QR code, use it as a watermark for validation of the authorship of the original code. The results of the stego images evaluated by means of the image quality metrics: SSIM, PSNR, UQI and VIF of the stego images and the recovery and decoding of the hidden QR images are presented.
. For many years, children's songs have been part of the growth of human beings, since they offer great benefits such as; the development of intelligence, the teaching of new values and
. Developing machine learning tools to aid students in the process of writing a thesis document is of great interest to students, universities, supervisors and evaluation committees. This article presents the construction and evaluation of readability comparators based in Spanish-written thesis documents of four different academic levels: Advanced College Level Technician (ACT), Undergraduate, Master and Doctoral. Specifically, we provide comparators that can evaluate, between two thesis texts which one is more readable than the other; the thesis sections we focus are: Problem Statement, Results and Justification. The successful completion of these different comparators, as shown in our results, opens the possibility for building a web-based API that analyzes an input thesis draft section and determines whether corresponds to its academic level or requires further improvement.
. In recent years, intentional homicides have increased drastically in Mexico. This increase of violence has only been analyzed considering descriptive statistical methods when they are applicable to the entire country
Factors such as global warming, pollution, overpopulation, and human unconsciousness are some of the causes that have generated a negative
Con la nueva oleada de las redes neuronales, arquitecturas basas en aprendizaje profundo han tenido mucho éxito en la última década. Una de las principales aplicaciones que han tenido estas arquitecturas es la generación automática de texto, la cual ha ganado popularidad recientemente. Distintas tareas han tratado de ser resueltas como generador de canciones, chatbots, resúmenes automáticos, traductores, entre otros. Sin embargo pocos trabajos se han enfocado en generar textos científicos. En este trabajo se presenta un análisis de generación de abstracts de artículos científicos explorando distintos tipos de arquitecturas. Para este trabajo se recolectaron de 227 artículos de Procesamiento de lenguaje natural aplicado al sector turístico. Con esta colección se proponen diferentes tipos de fine tuning donde el mejor resultado es de 0.21 obtenido por GPT-3 según el coeficiente de Jaccard.
. In the learning of the exact sciences, especially those related to engineering, there has been difficulty in achieving effective learning in the complex topics that these areas of knowledge include, affecting the motivational state of the students. Currently, some applications are based on the use of technology as a learning management tool. However, not all available tools are being used to better capture students' attention. This work aids with the learning of the topic "Electric circuits" in first-grade engineering students by promoting learning, understanding, and application of the benefits that augmented reality technology provides. In addition to being able to change content to each student's learning pace, the fuzzy logic technique based on student interaction allows for content adaptation.
Resumen. En la actualidad el tema de optimizaci´on en recursos na-turales es una preocupaci´on que compete a diferentes ´areas de la cien-cia. En espec´ıfico en lo que son los invernaderos o tambi´en conocidos como GreenHouse han sido una tendencia en la ´ultima d´ecada. Tanto tendencias de invernaderos en casa como a gran escala son de inter´es en la comunidad cient´ıfica. En este trabajo con el objetivo de adaptar una metodolog´ıa de control de invernaderos que pueda ser usado en un contexto local y regional se propuso un dise˜no basado en IoT y T´ecnicas de IA. Nuestra investigaci´on tuvo como parte de los objetivos la investigaci´on de diferentes dise˜nos de invernaderos, dise˜nos de riego, estudio de diferentes sensores y t´ecnicas de IA. Primero se hizo un modelo a escala y posteriormente se tiene proyectado aplicarlo a una escala en el Instituto Tecnol´ogico Superior de Pur´ısima del Rinc´on (ITSPR). En este trabajo reportamos la metodolog´ıa desempe˜nada y los resultados obtenidos del dise˜no propuesto.
