- Jose Crossa (3)
- ANA TERESA CELADA MURILLO (2)
- Andrew Chubykalo (2)
- BERNARDO FIGUEROA ESPINOZA (2)
- Carlo Montes (2)
- CICESE (9)
- Instituto Nacional de Astrofísica, Óptica y Electrónica (6)
- Universidad Autónoma Metropolitana (México). Unidad Azcapotzalco. Coordinación de Servicios de Información. (6)
- Universidad Autónoma de Baja California (5)
- CIATEQ, A.C. (3)
- Repositorio Institucional CICESE (17)
- Repositorio Institucional del INAOE (10)
- Repositorio Institucional de Publicaciones Multimedia del CIMMYT (8)
- CIATEQ Digital (7)
- Repositorio Institucional Zaloamati (6)
- oa:openAccess (71)
- CIENCIAS FÍSICO MATEMÁTICAS Y CIENCIAS DE LA TIERRA (33)
- INGENIERÍA Y TECNOLOGÍA (19)
- CIENCIAS TECNOLÓGICAS (16)
- CIENCIAS DE LA TIERRA Y DEL ESPACIO (15)
- OCEANOGRAFÍA (13)
Select the topics of your interest and receive the hottest publications in your email
This article presents a comparison of wind speed forecasting techniques, starting with the Auto-regressive Integrated Moving Average, followed by Artificial Intelligence-based techniques. The objective of this article is to compare these methods and provide readers with an idea of what method(s) to apply to solve their forecasting needs. The Artificial Intelligence-based techniques included in the comparison are Nearest Neighbors (the original method, and a version tuned by Differential Evolution), Fuzzy Forecasting, Artificial Neural Networks (designed and tuned by Genetic Algorithms), and Genetic Programming. These techniques were tested against twenty wind speed time series, obtained from Russian and Mexican weather stations, predicting the wind speed for 10 days, one day at a time. The results show that Nearest Neighbors using Differential Evolution outperforms the other methods. An idea this article delivers to the reader is: what part of the history of the time series to use as input to a forecaster? This question is answered by the reconstruction of phase space. Reconstruction methods approximate the phase space from the available data, yielding m (the system’s dimension) and τ (the sub-sampling constant), which can be used to determine the input for the different forecasting methods.
Long-term prediction of time series from chaotic systems is a difficult task but required in various fields such as economics, medicine, government, etc. In recent years, several studies have turned their attention to reuse prior knowledge with the aim of combining models and improving prediction. The process of learning from previous results is called meta-learning. On the context of model combination, the meta-learning strategy lets to automatically generate an expert, able to recommend an appropriate combination of models to predict a time series in particular. However, the process of meta-learning in time series imposes nontrivial challenges such as: analyzing model behavior with time series data, look for models that can be combined with each series and even generate new methods of meta-learning those consider variations of model performance over time. This research focuses on the problem of meta-learning of model combination, using self-organization and an automatic analysis of the behavior of the models. The goal is to improve the accuracy on long-term prediction achieved by state of the art algorithms in chaotic time series. The main contribution of this work is a new method, based on meta-features and self-organized maps neural networks, capable of generating combinations of prediction models considering the accuracy of the models and their diversity of behavior over time. The developed method was compared with other methods of the state of the art, and validated using synthetic series and series obtained from real applications, that have a chaotic and non-chaotic behavior.
La predicción a largo plazo de series de tiempo de sistemas caóticos es una tarea difícil pero requerida en diversos dominios como economía, medicina, gobierno, etc. En los últimos años, varias investigaciones han puesto su atención en reutilizar conocimiento previo a fin de combinar modelos y mejorar la predicción. El proceso de aprender a partir de resultados previos es denominado meta-aprendizaje. En el contexto de combinación de modelos, la estrategia de meta-aprendizaje permite generar automáticamente un experto, capaz de recomendar una combinación de modelos apropiada para predecir una serie de tiempo en particular. Sin embargo, el proceso de meta-aprendizaje en series de tiempo impone retos no triviales tales como: analizar el comportamiento de los modelos con los datos, buscar modelos que puedan combinarse adecuadamente para cada serie, e incluso generar nuevos métodos de meta-aprendizaje, que consideren las variaciones de desempeño de los modelos en el tiempo. Esta investigación se centra en el problema de meta-aprendizaje de combinación de modelos, utilizando auto-organización y un análisis automático del comportamiento de los modelos. La meta es mejorar la exactitud en predicción a largo plazo alcanzada por los algoritmos del estado del arte en series de tiempo caóticas. La principal contribución de este trabajo es un nuevo método, basado en meta-características y en las redes neuronales de mapas auto-organizados, capaz de generar combinaciones de modelos de predicción considerando la exactitud de los modelos y su diversidad de comportamiento en el tiempo. El método desarrollado fue comparado con otros métodos del estado del arte, y validado utilizando series sintéticas y series obtenidas de aplicaciones reales, que presentan comportamientos tanto caóticos como no caóticos.
ENRIQUE RIVERA CASTILLO (2012)
Tesis (Maestría en Ciencias, especialista en Estadística).- Colegio de Postgraduados, 2012.
La evapotranspiración de referencia (ETo) es un proceso no lineal empleado para determinar la cantidad de agua utilizada en los programas de irrigación. El nivel de precisión de esta variable a partir de datos históricos, ha sido siempre fundamental. En este trabajo, se presenta una aplicación de las Máquinas de Soporte Vectorial (SVMs) para la predicción de ETo y se compara su capacidad predictiva con otras dos metodologías de predicción: Redes Neuronales Artificiales de Multicapa (MLP) y modelos Autoregresivos Integrados de Promedio Móvil (ARIMA). Se propone un algoritmo heurístico de refinamiento para la implementación de las SVM resultando en una predicción mucho mejor que la obtenida con los otros dos métodos. La capacidad de predicción fue evaluada utilizando el Error Porcentual Medio Absoluto (MAPE). _______________ SUPPORT VECTOR MACHINES IN THE TIME SERIES ANALYSIS. ABSTRACT: Reference crop evapotranspiration (ETo) is a non linear process used to determine the quantity of water used in irrigation programs and the level of accuracy of the prediction of this variable from historical data has always been fundamental. In this work, we present an application of Support Vector Machines (SVMs) for ETo forecasting and compare its prediction capacity with two other prediction methodologies: Multi Layer Perceptron (MLP) neural networks and Auto-Regressive Integrated Moving Average (ARIMA) models. A proposed heuristic re nement algorithm for the implementation of the SVM gave a very good forecasting, much better than those obtained with the other two methods. Forecasting capacity was evaluated using the Mean Absolute Percentage Error (MAPE).
This paper reports the results of a research exploring the mathematical connections of pre-university students while they solving tasks which involving rates of change. We assume mathematical connections as a cognitive process through which a person finds real relationships between two or more ideas, concepts, definitions, theorems, procedures, representations or meanings or with other disciplines or the real-world. Four tasks were proposed to the 33 pre-university students that participated in this research; the central concept of the first task is the slope, the last three tasks contain conceptslikevelocity,speedandacceleration.Task-basedinterviews were conducted to collect data and later analysed with thematic analysis. Results showed most of the students made mathematical connections of the procedural type, the mathematical connections ofthecommonfeaturestypearemadeinsmallerquantitiesandthe mathematicalconnectionofthegeneralizationtypeisscarcelymade. Furthermore, students considered slope as a concept disconnected fromvelocity,speedandacceleration.
Genomic selection (GS) by selecting lines prior to field phenotyping using genotyping data has the potential to enhance the rate of genetic gains. Genotype × environment (G × E) interaction inclusion in GS models can improve prediction accuracy hence aid in selection of lines across target environments. Phenotypic data on 320 chickpea breeding lines for eight traits for three seasons at two locations were recorded. These lines were genotyped using DArTseq (1.6 K SNPs) and Genotyping-by-Sequencing (GBS; 89 K SNPs). Thirteen models were fitted including main effects of environment and lines, markers, and/or naïve and informed interactions to estimate prediction accuracies. Three cross-validation schemes mimicking real scenarios that breeders might encounter in the fields were considered to assess prediction accuracy of the models (CV2: incomplete field trials or sparse testing; CV1: newly developed lines; and CV0: untested environments). Maximum prediction accuracies for different traits and different models were observed with CV2. DArTseq performed better than GBS and the combined genotyping set (DArTseq and GBS) regardless of the cross validation scheme with most of the main effect marker and interaction models. Improvement of GS models and application of various genotyping platforms are key factors for obtaining accurate and precise prediction accuracies, leading to more precise selection of candidates.
Multi-environment trials are routinely conducted in plant breeding to select candidates for the next selection cycle. In this study, we compare the prediction accuracy of four developed genomic-enabled prediction models: (1) single-environment, main genotypic effect model (SM); (2) multi-environment, main genotypic effects model (MM); (3) multi-environment, single variance G×E deviation model (MDs); and (4) multi-environment, environment-specific variance G×E deviation model (MDe). Each of these four models were fitted using two kernel methods: a linear kernel Genomic Best Linear Unbiased Predictor, GBLUP (GB), and a nonlinear kernel Gaussian kernel (GK). The eight model-method combinations were applied to two extensive Brazilian maize data sets (HEL and USP data sets), having different numbers of maize hybrids evaluated in different environments for grain yield (GY), plant height (PH), and ear height (EH). Results show that the MDe and the MDs models fitted with the Gaussian kernel (MDe-GK, and MDs-GK) had the highest prediction accuracy. For GY in the HEL data set, the increase in prediction accuracy of SM-GK over SM-GB ranged from 9 to 32%. For the MM, MDs, and MDe models, the increase in prediction accuracy of GK over GB ranged from 9 to 49%. For GY in the USP data set, the increase in prediction accuracy of SM-GK over SM-GB ranged from 0 to 7%. For the MM, MDs, and MDe models, the increase in prediction accuracy of GK over GB ranged from 34 to 70%. For traits PH and EH, gains in prediction accuracy of models with GK compared to models with GB were smaller than those achieved in GY. Also, these gains in prediction accuracy decreased when a more difficult prediction problem was studied.
Hyperspectral cameras can provide reflectance data at hundreds of wavelengths. This information can be used to derive vegetation indices (VIs) that are correlated with agronomic and physiological traits. However, the data generated by hyperspectral cameras are richer than what can be summarized in a VI. Therefore, in this study, we examined whether prediction equations using hyperspectral image data can lead to better predictive performance for grain yield than what can be achieved using VIs. For hyperspectral prediction equations, we considered three estimation methods: ordinary least squares, partial least squares (a dimension reduction method), and a Bayesian shrinkage and variable selection procedure. We also examined the benefits of combining reflectance data collected at different time points. Data were generated by CIMMYT in 11 maize (Zea mays L.) yield trials conducted in 2014 under heat and drought stress. Our results indicate that using data from 62 bands leads to higher prediction accuracy than what can be achieved using individual VIs. Overall, the shrinkage and variable selection method was the best-performing one. Among the models using data from a single time point, the one using reflectance collected at 28 d after flowering gave the highest prediction accuracy. Combining image data collected at multiple time points led to an increase in prediction accuracy compared with using single-time-point data.
Investigadores del gobierno y la industria, anticipan el incremento en un futuro muy cercano de parques eólicos de baja velocidad de viento en Estados Unidos, Canadá, Europa, China, India y Brasil. Además, el fabricante de turbinas eólicas Siemens ha declarado que "espera que un tercio del desarrollo global de la energía eólica en los próximos años tenga lugar en áreas con velocidades del viento de medias a bajas". El escenario anterior bien puede aplicarse a cualquier parte del mundo, pues la búsqueda del aprovechamiento eólico no puede limitarse a rangos de velocidades de viento altos. Los vientos de media y baja velocidad, bien pueden explotarse con tecnología diseñada para las características propias de ese recurso específico. El presente trabajo, muestra el análisis estructural mediante el Método de Elemento Finito de un álabe de doble Raíz de baja capacidad para bajas velocidades de viento (patente en trámite ante el IMPI). El álabe, se considera fabricado con materiales compuestos. Para justificar el diseño, se consideran los casos de la metodología de carga simplificada contemplada en la norma IEC 61400-2. Los resultados arrojan evidencia sobre la viabilidad de estos álabes para su aplicación comercial.
Researchers from the government and industry are anticipating the increase in the near future of wind farms with low wind speeds in the United States, Canada, Europe, China, India and Brazil. In addition, the manufacturer of wind turbines Siemens has declared that "it expects that a third of the global development of wind energy in the coming years will take place in areas with medium to low wind speeds" . The previous scene can be applied to any part of the world, since the search for wind power can not be limited to ranges of high wind speeds. The medium and low speed winds can be exploited with technology designed for the specific characteristics of that specific resource. The present work, shows the structural analysis by means of the Method of Finite Element of a blade of double Root of low capacity for low speeds of wind (patent in proceeding before the IMPI). The blade is considered manufactured with composite materials. To justify the design, the cases of the simplified loading methodology contemplated in the IEC 61400-2 standard are considered. The results show evidence on the viability of these blades for commercial application.
We present two high-speed and low-power full-adder cells designed with an alternative internal logic structure and pass-transistor logic styles that lead to have a reduced power-delay product (PDP). We carried out a comparison against other full-adders reported as having a low PDP, in terms of speed, power consumption and area. All the full-adders were designed with a 0.18-μm CMOS technology, and were tested using a comprehensive testbench that allowed to measure the current taken from the full-adder inputs, besides the current provided from the power-supply. Post-layout simulations show that the proposed full-adders outperform its counterparts exhibiting an average PDP advantage of 80%, with only 40% of relative area.
Carlo Montes (2019)