Automatic detection of gait events: a case study using inductive learning techniques.

摘要: One of the problems which occurs in development a control system for functional electrical stimulation lower limbs is to detect accurately specific events within gait cycle. We present method classification phases cycle using artificial intelligence technique inductive learning. Both terminology learning and algorithm used analyses are fully explained. Given set examples sensor data from that be detected, able produce decision tree (or rules) classify minimum number sensors. The nature redundancy examined by progressively removing combinations sensors noting effect on both size trees produced their accuracy 'unseen' testing data. Since calculate more important (informative), comparisons with intuitive appreciation importance five researchers fields were made, revealing those appear intuitively most informative may, fact, provide least information. Comparison results standard statistical linear discriminant analysis also presented, showing relative simplicity inductively derived rules together good accuracy. In addition FES, such techniques applicable automatic construction expert systems diagnosis pathologies.