1 code implementation • 28 May 2020 • D. Gil, K. Díaz-Chito, C. Sánchez, A. Hernández-Sabaté
Results show that classic approaches can outperform deep-learning methods in this experimental setting, indicate the feasibility of early COVID-19 screening and that non-COVID infiltration is the group of patients most similar to COVID-19 in terms of radiological description of X-ray.
no code implementations • 18 Feb 2020 • O. Ramos Terrades, A. Berenguel, D. Gil
Outlier, or anomaly, detection is essential for optimal performance of machine learning methods and statistical predictive models.