Navegación

Búsqueda

Búsqueda avanzada

Towards a Fast and Accurate EIT Inverse Problem Solver: A Machine Learning Approach

Different industrial and medical situations require the non-invasive extraction of information from the inside of bodies. This is usually done through tomographic methods that generate images based oninternal body properties. However, the image reconstruction involves a mathematical inverse problem, which accurate resolution demands large computation time and capacity. In this paper we explore the use of Machine Learning to develop an accurate solver for reconstructing Electrical Impedance Tomography images on real-time. We compare the results with the Iterative Gauss-Newton and the Primal Dual Interior Point Method, which are both largely used and well-validated solvers. The approaches were compared from the qualitative as well as the quantitative viewpoints. The former was focused on correctly detecting the internal body features. The latter was based on accurately predicting internalproperty distributions. Experiments revealed that our approach achieved better accuracy and Cohen’s kappa coefficient (97.57% and 94.60% respectively) from the qualitative viewpoint. Moreover, it also obtained better quantitative metrics with a Mean Absolute Percentage Error of 18.28%. Experiments confirmed that Neural Networks algorithms can reconstruct internal body properties with high accuracy, so they would be able to replace more complex and slower alternatives.

TAPON: a two-phase machine learning approach for semantic labelling

Through semantic labelling we enrich structured information from sources such as HTML pages, tables, or JSON files, with labels to integrate it into a local ontology. This process involves measuring some features of the information and then finding the classes that best describe it. The problem with current techniques is that they do not model relationships between classes. Their features fall short when some classes have very similar structures or textual formats. In order to deal with this problem, we have devised TAPON: a new semantic labelling technique that computes novel features that take into account the relationships. TAPON computes these features by means of a two-phase approach. In the first phase, we compute simple features and obtain a preliminary set of labels (hints). In the second phase, we inject our novel features and obtain a refined set of labels. Our experimental results show that our technique, thanks to our rich feature catalogue and novel modelling, achieves higher accuracy than other state-of-the-art techniques.

Transformaciones de Datos con Machine Learning

Una de las tareas más comunes que los ingenieros tienen que llevar a cabo y que consumen más tiempo es la transformación de datos. Proponemos usar los avances en Inteligencia Artificial (IA), y en particular, en el área de Machine Learning (ML), para abordar este problema. Para ello, definimos una arquitectura que es capaz de inferir las transformaciones de datos a partir de un conjunto de pares de datos entrada-salida. Una vez que nuestro sistema haya aprendido cómo los datos de entrada se relacionan con los de salida, podrá realizar la traducción de nuevos datos de entrada automáticamente.

Un Recorrido por los Principales Proveedores de Servicios de Machine Learning y Predicción en la Nube

Los medios tecnológicos para el consumo, producción e intercambio de información no hacen más que aumentar cada día que pasa. Nos encontramos envueltos en el fenómeno Big Data, donde ser capaces de analizar esta informa- ción con el objetivo de poder inferir situaciones del futuro basándonos en datos del pasado y del presente, nos puede reportar una ventaja competitiva que nos distinga claramente de otras opciones. Dentro de las múltiples disciplinas exis- tentes para el análisis de grandes cantidades información encontramos el Ma- chine Learning y, a su vez, dentro de este podemos destacar la capacidad predic- tiva que nos proporcionan muchas de las opciones existentes actualmente en el mercado. En este trabajo realizamos un análisis de estas principales opciones de APIs predictivas en la nube, las comparamos entre sí, y finalmente llevamos a cabo una experimentación con datos reales de la Red de Vigilancia y Control de la Calidad del Aire de la Junta de Andalucía. Los resultados demuestran que estas herramientas son una opción muy interesante a considerar a la hora de tratar de predecir valores de contaminantes que pueden afectar a nuestra salud seriamente, pudiéndose llevar a cabo acciones preventivas sobre la población afectada.

Analysis of the Feasibility to Combine CEP and EDA with Machine Learning using the Example of Network Analysis and Surveillance

Complex Event Processing (CEP) and Event-driven Architectures (EDA) are modern paradigms for processing data in form of events. Machine Learning (ML) methods offer additional sophisticated means for analyzing data. By combining these technologies it is possible to create even more comprehensive and powerful data analysis and processing systems. We analyze the feasibility of combining CEP and EDA with ML using the example of the application domain of computer networks. We present relevant aspects, a sample use case, an sample architecture, and results of performance benchmarks. Our results indicate that the combination of these technologies increases data processing capabilities and that it is feasible from a performance perspective as well.