dc.contributorMontenegro Díaz, Álvaro Mauricio
dc.creatorCáliz Viñas, Arcesio Jose
dc.date.accessioned2022-11-08T19:05:10Z
dc.date.available2022-11-08T19:05:10Z
dc.date.created2022-11-08T19:05:10Z
dc.date.issued2022-08-01
dc.identifierhttps://repositorio.unal.edu.co/handle/unal/82665
dc.identifierUniversidad Nacional de Colombia
dc.identifierRepositorio Institucional Universidad Nacional de Colombia
dc.identifierhttps://repositorio.unal.edu.co/
dc.description.abstractIntelligent adaptive tests allow to perform evaluations that reduce the number of questions, improve the estimation, and adapt to the person’s answers. A decision to make in the design of these tests is the method for choosing the next question. There is no method that proves to be the best from the perspective of the improvements implied by an adaptive test. In this work, we explore the use of reinforcement learning algorithms in conjunction with deep learning algorithms for question choice. The results show that under certain conditions and depending on the algorithm used, these new methods achieve competent results in terms of the number of questions asked and the accuracy of the estimation compared to traditional statistical methods. (Texto tomado de la fuente)
dc.description.abstractLos test adaptativos inteligentes permiten realizar evaluaciones que reducen el número de preguntas, mejoran la estimación y se adaptan a las respuestas de la persona. Una decisión a tomar en el diseño de estas pruebas, es el método para escoger la siguiente pregunta. No existe un método que demuestre ser el mejor desde la perspectiva de las mejoras implicadas en un test adaptativo. En este trabajo exploramos el uso de algoritmos de aprendizaje por refuerzo en conjunto con algoritmos de aprendizaje profundo para la escogencia de las preguntas. Los resultados muestran que bajo ciertas condiciones y dependiendo del algoritmo utilizado, estos nuevos métodos logran resultados competentes en términos del número de preguntas hechas y la exactitud de la estimación comparándolos con los métodos estadísticos tradicionales.
dc.languageeng
dc.publisherUniversidad Nacional De Colombia
dc.publisherBogotá - Ciencias - Doctorado en Ciencias - Estadística
dc.publisherDepartamento de Estadística
dc.publisherFacultad de Ciencias
dc.publisherBogotá, Colombia
dc.publisherUniversidad Nacional de Colombia - Sede Bogotá
dc.relationRedCol
dc.relationLaReferencia
dc.relationApproximately Optimal Approximate Reinforcement Learning
dc.relationSpinning Up in Deep Reinforcement Learning
dc.relationTrust Region Policy Optimization
dc.relationA Global Information Approach to Computerized Adaptive Testing
dc.relationAn Introduction to Multivariate Statistical Analysis
dc.relationTestlet-Based Multidimensional Adaptive Testing
dc.relationComputerized Adaptive Testing: Theory and Practice
dc.relationGenerating Adaptive and Non-Adaptive Test Interfaces for Multidimensional Item Response Theory Applications
dc.relationTheory of Statistical Estimation
dc.relationPractical Methods of Optimization
dc.relationStatistical Inference
dc.relationA model for testing with multidimensional items
dc.relationDeep Reinforcement Learning with Double Q-learning
dc.relationHuman-level control through deep reinforcement learning
dc.relationLearning from Delayed Rewards
dc.relationReinforcement Learning: State-of-the-Art
dc.relationBatch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
dc.relationMultidimensional adaptive testing
dc.relationOn Information and Sufficiency
dc.relationMultidimensional adaptive testing with constraints on test content
dc.relationReinforcement Learning: An Introduction
dc.relationMultidimensional Adaptive Testing with Optimal Design Criteria for~Item Selection
dc.relationMultidimensional Item Response Theory
dc.relationDeep Reinforcement Learning in Action
dc.relationAI and Machine Learning for Coders
dc.relationmirt: A Multidimensional Item Response Theory Package for the R Environment
dc.relationSome latent trait models and their use in inferring an examinee's ability
dc.relationApplications of Item Response Theory to Practical Testing Problems
dc.relationLoglinear multidimensional IRT models for polytomously scored items
dc.rightsAtribución-NoComercial 4.0 Internacional
dc.rightshttp://creativecommons.org/licenses/by-nc/4.0/
dc.rightsinfo:eu-repo/semantics/openAccess
dc.titleIntelligent Adaptive Testing Using Machine Learning Techniques
dc.typeTrabajo de grado - Maestría


Este ítem pertenece a la siguiente institución