Publication: Sistema colaborativo para el etiquetado de datos
Loading...
Official URL
Full text at PDC
Publication Date
2022
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citations
Exportar
Abstract
Actualmente, al utilizar técnicas de aprendizaje automático es necesario pasar por una etapa de etiquetado manual de datos para entrenar al software. Este proceso de etiquetado puede resultar muy tedioso y ocupar una gran cantidad de tiempo porque en muchas ocasiones debe realizarse de manera manual.
Este proyecto consiste en implementar un sistema colaborativo para etiquetar datos de forma más eficiente y fiable. La estrategia que hemos decidido para desarrollar este sistema es implementar un servicio de CAPTCHA. Este tipo de servicios suele funcionar presentando un reto que normalmente consiste en seleccionar los elementos dentro de un conjunto de datos que cumplen una condición determinada, por lo que las respuestas que envían los usuarios que realizan estos retos pueden servir para clasificar esos elementos que han seleccionado. Sin darse cuenta, los usuarios a los que el reto haya verificado como reales, están realizando un proceso de etiquetado de datos.
Currently, when using machine learning techniques, it is necessary to go through a stage of manual data labeling to train the software. This data labeling process can be very tedious and take a lot of time because often it is done manually. This project consists of implementing a collaborative system to label data more efficiently and reliably. The strategy we have decided to develop this system is to implement a CAPTCHA service. This type of service usually works by presenting a challenge that generally consists of selecting the elements within a set of data that meet a certain condition, so the answers sent by users who perform these challenges can be used to classify those elements they have selected. Without being aware, users who have been verified as real by the challenge are performing a data labeling process.
Currently, when using machine learning techniques, it is necessary to go through a stage of manual data labeling to train the software. This data labeling process can be very tedious and take a lot of time because often it is done manually. This project consists of implementing a collaborative system to label data more efficiently and reliably. The strategy we have decided to develop this system is to implement a CAPTCHA service. This type of service usually works by presenting a challenge that generally consists of selecting the elements within a set of data that meet a certain condition, so the answers sent by users who perform these challenges can be used to classify those elements they have selected. Without being aware, users who have been verified as real by the challenge are performing a data labeling process.
Description
Trabajo de Fin de Grado en Ingeniería Informática, Departamento de Sistemas Informáticos y Computación, Facultad de Informatica UCM, Curso 2021/2022.
Todos los recursos del proyecto están disponibles en el siguiente enlace: https://github.com/AdriSC/tfg