A5080161418- Machine Learning and Data Classification
- Mobile Crowdsensing and Crowdsourcing
- Data Stream Mining Techniques
- Machine Learning and Algorithms
- Imbalanced Data Classification Techniques
- Bayesian Modeling and Causal Inference
- Text and Document Classification Technologies
- Mental Health Research Topics
- Ovarian function and disorders
- Artificial Intelligence in Healthcare and Education
- Reproductive Biology and Fertility
- Software Engineering Research
- Health, Environment, Cognitive Aging
- Resilience and Mental Health
- Data Management and Algorithms
- Machine Learning in Healthcare
- Health disparities and outcomes
- Microplastics and Plastic Pollution
- Artificial Intelligence in Healthcare
- Bayesian Methods and Mixture Models
- Gaussian Processes and Bayesian Inference
- Chaos control and synchronization
- Natural Language Processing Techniques
- Fault Detection and Control Systems
- Nonlinear Dynamics and Pattern Formation
Universitat de Girona
2023-2025
Universitat de Barcelona
2020-2023
University of the Basque Country
2011-2019
Artificial Intelligence Research Institute
2019
Consejo Superior de Investigaciones Científicas
2019
Universidad Politécnica Metropolitana de Hidalgo
2018
Ideko (Spain)
2006
A significant level of stigma and inequality exists in mental healthcare, especially under-served populations. Inequalities are reflected the data collected for scientific purposes. When not properly accounted for, machine learning (ML) models learned from can reinforce these structural inequalities or biases. Here, we present a systematic study bias ML designed to predict depression four different case studies covering countries We find that standard approaches regularly biased behaviors....
OBJECTIVES: To assess the value of machine learning approaches in development a multivariable model for early prediction ICU death patients with acute respiratory distress syndrome (ARDS). DESIGN: A development, testing, and external validation study using clinical data from four prospective, multicenter, observational cohorts. SETTING: network multidisciplinary ICUs. PATIENTS: total 1,303 moderate-to-severe ARDS managed lung-protective ventilation. INTERVENTIONS: None. MEASUREMENTS AND MAIN...
Machine learning techniques have been previously used to assist clinicians select embryos for human-assisted reproduction. This work aims show how an appropriate modeling of the problem can contribute improve machine embryo selection. In this study, a dataset 330 consecutive cycles (and associated embryos) carried out by Unit Assisted Reproduction Hospital Donostia (Spain) throughout 18 months has analyzed. The selection modeled novel weakly supervised paradigm, from label proportions, which...
Floods affect chronically many communities around the world. Their socioeconomic impact increases year-by-year, boosted by global warming and climate change. Combined with long-term preemptive measures, preparatory actions are crucial when floods imminent. In last decade, machine learning models have been used to anticipate these hazards. this work, we model Ter river (NE Spain), which has historically suffered from floods, using traditional ML such as K-nearest neighbors, Random forests,...
<title>Abstract</title> Mental illnesses affect almost 15% of the world's population, with half cases emerging before age 14. Improved methods for predicting progression mental distress among adolescents, particularly in vulnerable populations, are needed. This study utilized traditional machine learning techniques to predict health status at 17. We assessed correlates outcomes a sample 632 adolescents general (i.e., total difficulties score 17 or higher) 11, who participated UK Millennium...
Majority voting is a popular and robust strategy to aggregate different opinions in learning from crowds, where each worker labels examples according their own criteria. Although it has been extensively studied the binary case, its behavior with multiple classes not completely clear, specifically when annotations are biased. This paper attempts fill that gap. The of majority in-depth multi-class domains, emphasizing effect annotation bias. By means complete experimental setting, we show...
Learning from crowds is a classification problem where the provided training instances are labeled by multiple (usually conflicting) annotators. In different scenarios of this problem, straightforward strategies show an astonishing performance. paper, we characterize crowd these basic good behavior. As consequence, study allows to identify those non-basic methods for combining labels expected obtain better results. context, extend learning paradigm multidimensional (MD) domain. Measuring...
In peer assessment, students assess a task done by their peers, provide feedback and usually grade. The extent to which these grades can be used formally grade the is unclear, with doubts often arising regarding validity. instructor could supervise assessments, but would not then benefit from workload reduction, one of most appealing features assessment for instructors. Our proposal uses probabilistic model estimate each test, accounting degree precision bias grading peers. that assign test...
Weakly supervised classification tries to learn from data sets which are not certainly labeled. Many problems, with different natures of partial labeling, fit this description. In paper, the novel problem learning positive-unlabeled proportions is presented. The provided examples unlabeled, and only class information available consists positive unlabeled in subsets training set. We present a methodology that adapts levels uncertainty Bayesian network classifiers using an...
Over the last decade, hundreds of thousands volunteers have contributed to science by collecting or analyzing data. This public participation in science, also known as citizen has significant discoveries and led publications major scientific journals. However, little attention been paid data quality issues. In this work we argue that being able determine accuracy obtained crowdsourcing is a fundamental question point out that, for many real-life scenarios, mathematical tools processes...
Classifying software defects according to any defined taxonomy is not straightforward. In order be used for automatizing the classification of defects, two sets defect reports were collected from public issue tracking systems different real domains. Due lack a domain expert, categorized by set annotators unknown reliability their impact IBM's orthogonal taxonomy. Both datasets are prepared solve problem means techniques learning crowds paradigm (Hernández-González et al. [1]). Two versions...
There exist some physical systems whose underactuation and limited sensing represent a challenge in engineering. The former requires model-based nesting or virtual control schemes to the underactuated degrees of freedom through actuated ones, while latter imposes stringent requirements design controller using only partial access state, called output feedback. In this paper, we address feedback class uncertain nonlinear that is not neither require state estimation. Output are circumvented by...
Embryo selection is a critical step in assisted reproduction: good criteria are expected to increase the probability of inducing pregnancy. Machine learning techniques have been applied for implantation prediction or embryo quality assessment, which embryologists can use make decision about selection. However, this highly uncertain real-world problem, and current proposals do not model always all sources uncertainty. We present novel probabilistic graphical that accounts three different...