MACHINE LEARNING IN CONTROL OF FUNCTIONAL ELECTRICAL-STIMULATION SYSTEMS FOR LOCOMOTION

Two machine learning techniques were evaluated for automatic design of a rule-based control of functional electrical stimulation (FES) for l ocomotion of spinal cord injured humans, The task was to learn the inv ariant characteristics of the relationship between sensory information and the FES-control signal by using off-line supervised training, Sen sory signals were recorded using pressure sensors installed in the ins oles of a subject's shoes and goniometers attached across the joints o f the affected leg, The ITS-control consisted of pulses corresponding to time intervals when the subject pressed on the manual pushbutton to deliver the stimulation during FES-assisted ambulation, The machine l earning techniques used were the adaptive logic network (ALN) [1] and the inductive learning algorithm (IL) [2], Results to date suggest tha t, given the same training data, the IL learned faster than the ALN, w hile both performed the test rapidly. The generalization was estimated by measuring the test errors and it was better with an ALN, especiall y if past points were used to reflect the time dimension, Both techniq ues were able to predict future stimulation events, An advantage of th e ALN over the IL was that ALN's can be retrained with new data withou t losing previously collected knowledge, The advantages of the IL over the ALN were that the IL produces small, explicit, comprehensible tre es and that the relative importance of each sensory contribution can b e quantified.