Examinando por Materia "Random forest classifier"

Mostrando 1 - 3 de 3
  • No hay miniatura disponible
    Ítem
    Cerebral Cortex Atlas of Emotional States Through EEG Processing
    (SPRINGER, 2019-10-14) Gómez A.; Quintero O.L.; Lopez-Celani N.; Villa L.F.; Universidad EAFIT. Escuela de Ciencias; Modelado Matemático
    This paper addresses the cerebral cortex maps construction from EEG signals getting an information simplification method for an emotional state phenomenon description. Bi-dimensional density distribution of main signal features are identified and a comparison to a previous approach is presented. Feature extraction scheme is performed via windowed EEG signals Stationary Wavelet Transform with the Daubechies Family (1–10); nine temporal and spectral descriptors are computed from the decomposed signal. Recursive feature selection method based on training a Random forest classifier using a one-vs-all scheme with the full features space, then a ranking procedure via gini importance, eliminating the bottom features and restarting the entire process over the new subset. Stopping criteria is the maximum accuracy. The main contribution is the analysis of the resulting subset features as a proxy for cerebral cortex maps looking for the cognitive processes understanding from surface signals. Identifying the common location of different emotional states in the central and frontal lobes, allowing to be strong parietal and temporal lobes differentiators for different emotions. © 2020, Springer Nature Switzerland AG.
  • No hay miniatura disponible
    Ítem
    Spatial layout and surface reconstruction from omnidirectional images
    (Institute of Electrical and Electronics Engineers Inc., 2016-01-01) Posada, L.F.; Velasquez-Lopez, A.; Posada, L.F.; Velasquez-Lopez, A.; Universidad EAFIT. Departamento de Ciencias; Ciencias Biológicas y Bioprocesos (CIBIOP)
    This paper presents a spatial layout recovery approach from single omnidirectional images. Vertical structures in the scene are extracted via classification from heterogeneous features computed at the superpixel level. Vertical surfaces are further classified according to their main orientation by fusing oriented line features, floor-wall boundary features and histogram of oriented gradients (HOG) with a Random Forest classifier. Oriented line features are used to build an orientation map that considers the main vanishing points. The floor-wall boundary feature attempts to reconstruct the scene shape as if it were observed from a bird's-eye view. Finally, the HOG descriptors are aggregated per superpixel and summarize the gradient distribution at homogeneous appearance regions. Compared to existing methods in the literature which rely only on corners or lines, our method gains statistical support from multiple cues aggregated per superpixel which provide more robustness against noise, occlusion, and clutter. © 2016 IEEE.
  • No hay miniatura disponible
    Ítem
    Spatial layout and surface reconstruction from omnidirectional images
    (Institute of Electrical and Electronics Engineers Inc., 2016-01-01) Posada, L.F.; Velasquez-Lopez, A.
    This paper presents a spatial layout recovery approach from single omnidirectional images. Vertical structures in the scene are extracted via classification from heterogeneous features computed at the superpixel level. Vertical surfaces are further classified according to their main orientation by fusing oriented line features, floor-wall boundary features and histogram of oriented gradients (HOG) with a Random Forest classifier. Oriented line features are used to build an orientation map that considers the main vanishing points. The floor-wall boundary feature attempts to reconstruct the scene shape as if it were observed from a bird's-eye view. Finally, the HOG descriptors are aggregated per superpixel and summarize the gradient distribution at homogeneous appearance regions. Compared to existing methods in the literature which rely only on corners or lines, our method gains statistical support from multiple cues aggregated per superpixel which provide more robustness against noise, occlusion, and clutter. © 2016 IEEE.