VASS: herramienta docente web para la visualización y enseñanza de algoritmos de aprendizaje semisupervisado

Abstract

This article presents the VASS teaching tool, a web application designed to facilitate the teaching of semisupervised learning algorithms, a relatively new technique in machine learning that, in addition to labeled data, uses unlabeled data to improve the performance of machine learning models. This is especially necessary in those contexts where the acquisition of labeled data is laborious or very expensive. VASS (Visualizer of semisupervised Algorithms) has an intuitive interface that allows users to train and visualize the inner workings of four key semisupervised algorithms: Self-Training, Co-Training, TriTraining and Democratic Co-Learning. The application has been developed with its usefulness in educational environments in mind, providing students and teachers with a valuable tool to explore and understand these fundamental concepts. VASS not only seeks to improve the accessibility of semisupervised algorithms, but also to foster a deeper understanding of their functionality