GRIAL resources
Permanent URI for this communityhttps://repositorio.grial.eu/handle/123456789/1
Browse
6 results
Search Results
Item Testing and Improvements of KoopaML: A Platform to Ease the Development of Machine Learning Pipelines in the Medical Domain(Springer, 2023-05-01) Vázquez-Ingelmo, A.; García-Holgado, A.; García-Peñalvo, F. J.; Fraile-Sanchón, R.; Pérez-Sánchez, P.; Antúnez-Muiños, P.; Sánchez-Puente, A.; Vicente-Palacios, V.; Dorado-Díaz, P. I.; Cruz-González, I.; Sánchez, P. L.Machine Learning (ML) applications in complex domains, such as the medical domain, can be highly beneficial, but also hazardous if some concepts are overlooked. In this context, however, health professionals denote expertise in their domain, but they often lack skills in terms of ML. In this sense, to leverage ML applications in the medical domain, it is important to combine both domain expertise and ML-related skills. In previous works, we tackled this challenge in the health context through a visual platform (KoopaML) that enables lay users to build ML pipelines. The present work describes the challenges derived from the first version of the platform and the prototypes for the new features designed to address them. The prototypes have been validated by two experts, obtaining highly valuable feedback.Item What Do We Mean by GenAI? A Systematic Mapping of The Evolution, Trends, and Techniques Involved in Generative AI(2023-08-01) García-Peñalvo, F. J.; Vázquez-Ingelmo, A.Artificial Intelligence has become a focal point of interest across various sectors due to its ability to generate creative and realistic outputs. A specific subset, generative artificial intelligence, has seen significant growth, particularly in late 2022. Tools like ChatGPT, Dall-E, or Midjourney have democratized access to Large Language Models, enabling the creation of human-like content. However, the concept 'Generative Artificial Intelligence' lacks a universally accepted definition, leading to potential misunderstandings. While a model that produces any output can be technically seen as generative, the Artificial Intelligent research community often reserves the term for complex models that generate high-quality, human-like material. This paper presents a literature mapping of AI-driven content generation, analyzing 631 solutions published over the last five years to better understand and characterize the Generative Artificial Intelligence landscape. Our findings suggest a dichotomy in the understanding and application of the term "Generative AI". While the broader public often interprets "Generative AI" as AI-driven creation of tangible content, the AI research community mainly discusses generative implementations with an emphasis on the models in use, without explicitly categorizing their work under the term "Generative AI".Item Aplicación para la recepción, almacenamiento y gestión de imágenes DICOM en el sector sanitario(Grupo GRIAL, 2023-01-24) Fraile Sanchón, R.DICOM significa “Imágenes Digitales y Comunicaciones en Medicina” y fue desarrollado conjuntamente por la Asociación Nacional de Fabricante (NEMA) y el Colegio Americano de Radiología (ACR) para permitir la interoperabilidad entre equipos de imágenes con otros dispositivos. Este estándar es responsable de gobernar tanto el formato de imagen como los diversos protocolos de red necesarios para la transmisión de información de imágenes generadas durante las muchas “modalidades” de imágenes relacionadas con la atención médica, tales como resonancia magnética, medicina nuclear, tomografía computarizada y ultrasonidos. Por lo tanto, el estándar DICOM existe de una forma u otra desde 1983 y continúa evolucionando cada año. En el marco del manejo y gestión de imágenes DICOM, este proyecto consiste en la ampliación de una aplicación ya existente como es KoopaML, aplicación web enmarcada en el contexto del Departamento de Cardiología del Hospital Universitario de Salamanca, cuyo objetivo es permitir que sus usuarios puedan entrenar sus propios modelos, analizar sus datos y realizar tareas sobre ellos sin la necesidad de tener conocimientos de programación. A esta herramienta se le suma la existencia de Cartier IA, plataforma de almacenamiento y visualización de datos e imágenes médicas usada también en el Departamento de Cardiología. Esta ampliación se basa en la recepción, almacenamiento y gestión de imágenes DICOM en KoopaML. La recepción y almacenamiento se realiza a través de la transmisión de estos archivos a través de otros PACS (Sistema de Archivo y Comunicación de Imágenes) como puede ser Cartier IA, de esta manera se evita que los usuarios de ambos sistemas tengan que realizar unas acciones en un sistema y otras en el otro y así ahorrar tiempo. En cuanto a la gestión de estas imágenes, se les pueden aplicar algoritmos de Inteligencia Artificial a éstas y, modificarlas utilizando diferentes herramientas de edición para medir, anotar, recortar, acercar, desplazar y segmentar entre otras.Item Advances in the use of domain engineering to support feature identification and generation of information visualizations(ACM, 2020-10-21) Vázquez-Ingelmo, A.; García-Peñalvo, F. J.; Therón, R.Information visualization tools are widely used to better understand large and complex datasets. However, to make the most out of them, it is necessary to rely on proper designs that consider not only the data to be displayed, but also the audience and the context. There are tools that already allow users to configure their displays without requiring programming skills, but this research project aims at exploring the automatic generation of information visualizations and dashboards in order to avoid the configuration process, and select the most suitable features of these tools taking into account their contexts. To address this problem, a domain engineering, and machine learning approach is proposed.Item On data-driven systems analyzing, supporting and enhancing users’ interaction and experience(Grupo GRIAL, 2018-09-03) Cruz-Benito, J.The research areas of Human-Computer Interaction and Software Architectures have been traditionally treated separately, but in the literature, many authors made efforts to merge them to build better software systems. One of the common gaps between software engineering and usability is the lack of strategies to apply usability principles in the initial design of software architectures. Including these principles since the early phases of software design would help to avoid later architectural changes to include user experience requirements. The combination of both fields (software architectures and Human-Computer Interaction) would contribute to building better interactive software that should include the best from both the systems and user-centered designs. In that combination, the software architectures should enclose the fundamental structure and ideas of the system to offer the desired quality based on sound design decisions. Moreover, the information kept within a system is an opportunity to extract knowledge about the system itself, its components, the software included, the users or the interaction occurring inside. The knowledge gained from the information generated in a software environment can be used to improve the system itself, its software, the users’ experience, and the results. So, the combination of the areas of Knowledge Discovery and Human-Computer Interaction offers ideal conditions to address Human-Computer-Interaction-related challenges. The Human-Computer Interaction focuses on human intelligence, the Knowledge Discovery in computational intelligence, and the combination of both can raise the support of human intelligence with machine intelligence to discover new insights in a world crowded of data. This Ph.D. Thesis deals with these kinds of challenges: how approaches like data-driven software architectures (using Knowledge Discovery techniques) can help to improve the users' interaction and experience within an interactive system. Specifically, it deals with how to improve the human-computer interaction processes of different kind of stakeholders to improve different aspects such as the user experience or the easiness to accomplish a specific task. Several research actions and experiments support this investigation. These research actions included performing a systematic literature review and mapping of the literature that was aimed at finding how the software architectures in the literature have been used to support, analyze or enhance the human-computer interaction. Also, the actions included work on four different research scenarios that presented common challenges in the Human-Computer Interaction knowledge area. The case studies that fit into the scenarios selected were chosen based on the Human-Computer Interaction challenges they present, and on the authors’ accessibility to them. The four case studies were: an educational laboratory virtual world, a Massive Open Online Course and the social networks where the students discuss and learn, a system that includes very large web forms, and an environment where programmers develop code in the context of quantum computing. The development of the experiences involved the review of more than 2700 papers (only in the literature review phase), the analysis of the interaction of 6000 users in four different contexts or the analysis of 500,000 quantum computing programs. As outcomes from the experiences, some solutions are presented regarding the minimal software artifacts to include in software architectures, the behavior they should exhibit, the features desired in the extended software architecture, some analytic workflows and approaches to use, or the different kinds of feedback needed to reinforce the users’ interaction and experience. The results achieved led to the conclusion that, despite this is not a standard practice in the literature, the software environments should embrace Knowledge Discovery and data-driven principles to analyze and respond appropriately to the users’ needs and improve or support the interaction. To adopt Knowledge Discovery and data-driven principles, the software environments need to extend their software architectures to cover also the challenges related to Human-Computer Interaction. Finally, to tackle the current challenges related to the users’ interaction and experience and aiming to automate the software response to users’ actions, desires, and behaviors, the interactive systems should also include intelligent behaviors through embracing the Artificial Intelligence procedures and techniques.Item Proposing a machine learning approach to analyze and predict employment and its factors(2018-08-28) García-Peñalvo, F. J.; Cruz-Benito, J.; Martín-González, M.; Vázquez-Ingelmo, A.; Sánchez-Prieto, J. C.; TherónThis paper presents an original study with the aim of propose and test a machine learning approach to research about employability and employment. To understand how the graduates get employed, researchers propose to build predictive models using machine learning algorithms, extracting after that the most relevant factors that describe the model and employing further analysis techniques like clustering to get deeper insights. To test the proposal, is presented a case study that involves data from the Spanish Observatory for Employability and Employment (OEEU). Using data from this project (information about 3000 students), has been built predictive models that define how these students get a job after finalizing their degrees. The results obtained in this case study are very promising, and encourage authors to refine the process and validate it in further research.