Proyectos

En el Instituto de Informática, nuestros proyectos son el reflejo de nuestra dedicación a la investigación aplicada y la innovación. Trabajamos en iniciativas que buscan resolver problemas reales y generar un impacto positivo en la sociedad, a través de soluciones tecnológicas y colaborativas. Estos proyectos son el resultado del esfuerzo conjunto de académicos, investigadores y estudiantes, quienes abordan desafíos complejos y contribuyen al desarrollo de nuevas tecnologías, siempre con el objetivo de avanzar hacia un futuro más eficiente y sostenible.

Proyecto 1
Investigador responsable: Eliana Scheihing.

Áreas de investigación:
Data Science, Learning Analytics, Educational Data Mining, Bayesian Statistics, Visualization and Social Web Tools.

Descripción:
Actualmente colaboro con Valeria Henríquez, Daniel Guerra, Mauricio Ruiz-Tagle y académicos de la FFHH en la conformación de un núcleo de investigación para la transformación educativa desde el enfoque de ingeniería del aprendizaje. 
Investigador responsable: Matthieu Vernier.

Áreas de investigación:
Computational Social Science, Data Science, Natural Language Processing, Computational Linguistics, Critical Thinking.
 
Descripción: My project consists to create and evaluate computer-assisted protocols that improve Critical Thinking of young citizens towards Media Discourse using a methodology based on Data Science and Serious Games. In parallel, i am exploring data stream mining techniques to describe Concept-Drift in natural language streams. In other words, how the distribution of concepts and words is changing through time and to what extent it has an impact on supervised text classification models. This issue has many possible applications, one that i am currently exploring is the understanding of media discourse evolution in time. I develop this research with the following partners: UACh (teams: Informatics, Social Comunication, Neuroscience), University of Caen-France (team: Hultech-GREYC) and Universidad de Santiago de Compostela-Spain (team: Novos Medio). – From a more industrial perspective, i am collaborating with some start-up companies in Chile and France that are using data to explore social and cognitive phenomena to produce value for B2C companies (one of these startup is from Valdivia: ‘Spike’).
Investigador responsable: Luis Álvarez.

Áreas de investigación:
Tecnologías de Aprendizaje, Analítica de Aprendizaje, Aprendizaje personalizado.
 
Descripción: Junto a Valeria Henríquez, tenemos el proyecto “FIC21-31. Aprender y Enseñar en el siglo XXI”, que consiste en metodologías y plataformas para el aprendizaje de las matemáticas y programación en estudiantes de establecimientos públicos de la Región de los Ríos

Investigador responsable: Cristóbal Navarro.

Áreas de investigación: High Performance Computing
Computer Graphics
Computational Physics.

Descripción: Typical GPU computing patterns include nearest neighbors computations, all-pairs interactions, convolutions, tiled simulations, reductions, among others. Several of these patterns are used in applications for natural sciences, simulations, data mining, networks, interactions, social sciences, videogames, etc. A great motivation to research on this topic is the fact that a contribution on improving one of these GPU computation patterns can have a positive impact on families of application problems. Currently I am researching on answering the question of how much a GPU computing pattern ,that works on an $m$-simplex, can improve if we design an efficient thread map?

Investigador responsable: Cristóbal Navarro.

Áreas de investigación: High Performance Computing
Computer Graphics
Computational Physics.

Descripción: This research is a new line I am highly interested in developing. I plan to study how can we further improve the performance of modern ray tracing algorithms that use path tracing + machine learning. Machine learning contributed greatly at achieving real time ray tracing. However, today it is still an expensive computation as each frame applies a small number of simulation steps followed by a denoising process that it is still expensive enough to make consumer hardware unable to run videogames in real-time when using ray tracing. An important research question is what performance improvements can be done to the actual model of Machine Learning + Path Tracing used for real time ray tracing?

Investigador responsable: Cristian Olivares Rodríguez.

Áreas de investigación:
Educacional Data Mining, Learning Analytics and Machine Learning. Mathematical, computational and creative thinking.

Descripción: Nowadays, I’m developing a k-12 student behaviour analysis based on both clickstream and queries in order to acquire a better understood regarding to computational and creative thinking. (1) At computational thinking, I have analized the behaviour of a huge amount of students which are solved at most a set of 15 computational problems. Such data was acquired by using a block-based programming interface (www.kodetu.org) and it describes the solving process based on errors, times, (partial) solutions, and other fine graned data. However, there are not user models which predict or classify the user failures or problem dificulty, in such a way to support the learning process of computational thinking. (2) At creative thinking, I have developed a set of user models which are able to predict the creative quality of solutions based on query pattern, However, such information is not enough to obtain rich descriptions and predictions about how creative students are. In such a way, I’m interested in to improve the models by using detailed trace data about search information over several kind of problems. (3) Finally, I’m very interested in to develop a formal model about how computational and creative thinking are related by using both a data driven and a conceptual analysis.
Investigador responsable: Christian Lazo.

Áreas de investigación:
iot, sistemas inteligentes.

Descripción:
Desarrollo de tecnologías hardware y software.
Investigador responsable: Pablo Huijse.

Áreas de investigación: Machine Learning
Information Theory
Astroinformatics and Astrostatistics.

Descripción:
Astronomy has become an example of intensive data-driven science. Near future surveys will issue millions of nightly alerts with data rates bordering 1PB per night. This is particularly important for Chile because by 2020 70% of the astronomical observing capacity will be installed in our country. New interdisciplinary sciences such as astroinformatics and astrostatistics have been created to deal with these challenges. My research is on autonomous and efficient methods to discriminate and classify data from astronomical surveys. I develop scalable machine learning methods with solid information theoretic background to analyze astronomical time series, images, and astronomical transient alerts. The objectives are to classify phenomena corresponding to one of the currently known astrophysical categories and to detect novel phenomena arising in the data. We are also developing an astronomical alert broker (www.alerce.science) to deal with the deluge of data from the Large Synoptic Survey Telescope (LSST).
Investigador responsable: Héctor Ferrada.

Áreas de investigación:
High Performing Algorithms
Data Science
Compact Data Structures
Data Compression
Repetitive Large-Scale Data.

Descripción:
I am focusing in design new High Performing Algorithms to build Compact Data Structures in the fields of Data Science and Biological Sequences Analysis/Processing, addressing problems with large-scale data storage.
One of the current problems to do Data Science in Big Data is to consider the complete input data in its analysis, which generates big problems given the amount of data. The new idea is, first, to preprocess the input data with modern data compression techniques, using High Performance Computing, to exploit the data regularities in order to achieve compression; and second, to build small representations of the input data that can fit in main memory, which additionally offer efficient information retrieval capabilities, as pattern searches and other useful functionalities, in order to facility a deep analisys of them.
Investigador responsable: Julio Daniel Guerra H.

Áreas de investigación: Open Learner Models
Learning Analytics
Educational Data Mining
Self-Regulated Learning and Learning Motivation.

Descripción:
My areas of research involve educational technologies, adaptive educational systems, educational data mining, visualization and learning analytics. The analyses of learning and academic data is challenged by the fact that learning is a complex process that depends in cross-dependent variables involving attitudinal, motivational, cognitive and metacognitive aspects. In this scenario, my work focused in using artificial intelligence and statistical tools to develop and deploy learning interventions, systems and studies.
Investigador responsable: Valeria Henríquez N.

Áreas de investigación: Agile Software Development
Agile Software Process Improvemen Software Engineering Education & Training
Technology Enhanced Learning.


Descripción: Mi investigación se ha enfocado en dos principales área, la ingeniería de software y la tecnología y datos como potenciadores de la educación. En el área de ingeniería de software, me he enfocado en la implementación de métodos ágiles a nivel organizacional en empresas de desarrollo de software, con el objetivo de mejorar su desempeño. Esto incluye la adopción de modelos de mejora continua, la implementación de prácticas de calidad de software y la gestión de proyectos de software de manera ágil.
Investigador responsable: Luis Veas.

Áreas de investigación: Simulación Computacional, Computación Paralela y Distribuida, Base de Datos SQL / NoSQL, Sistemas Basados en la Web.

Descripción: Actualmente, mi investigación se centra en el uso de la WEB como puente para generar sistemas de utilidad para las personas o comunidades y el uso masivo de información. Actualmente, estoy trabajando en varios proyectos, donde los tres más relevantes son: la Plataforma GEOParques, la Plataforma PlaFarma y FuSA Roads. La Plataforma GEOParques, es un ecosistema multiperspectiva que permite a turistas acceder a información interesante de forma georeferenciada, permitiéndole planificar sus visitas, realizar visitas autoguiadas y mejorando su experiencia. La Plataforma PlaFarma surge como producto de la investigación conjunta y multidisciplinaria realizada por los integrantes de los núcleos de I+D HASFa y RINHoS, de los cuales formo parte en UACh. Los objetivos de PlaFarma son dos, el primero, la generación y disponibilización de algoritmos de farmacovigilancia, para alertar de medicamentos potencialmente riesgosos en prescripciones médicas, el segundo, generar un dashboard u observatorio de datos para permitir la exploración sobre los datos de 6 años de prescripciones médicas y de esta forma ayudar a la toma de decisiones informadas, por parte de las autoridades de salud. El proyecto FuSA Roads, busca generar mapas de ruido ambiental basado en los registros de estaciones de monitoreo vehicular, y haciendo uso del video y audio registrados en estas estaciones.