The SB-ST decomposition in the study of Developmental Coordination Disorder

摘要: To deal with redundancy and high-dimensionality that are typical of movement data, we propose to decompose action matrices in two decoupled steps: first, discover a set key postures, is, vectors corresponding relationships among degrees freedom (like angles between body parts) which call spatial basis (SB) second, impose parametric model the spatio-temporal (ST) profiles each SB vector. These steps constitute SB-ST decomposition an action: represent their ST trajectories these postures parameters express how being controlled coordinated. shares elements common computational models motor synergies biological motion perception, it relates human manifold popular machine learning. We showcase method by applying study vertical jumps adults, typically developing children Developmental Coordination Disorder obtained capture. Using also evaluate alone against other techniques terms reconstruction ability number dimensions used.