Classification of nerve signals using Kohonen's self-organizing map
Mariusz Bogdan · 1996
: In this paper we shortly review the aim of the INTER 1 -project (Intelligent Neural InTERface) and present our proposal for the modus operandi to process real nerve signals. In the second part we present a SelfOrganizing Map (SOM) that classify nerve signals from the dorsal ventricular nerve of a crab. INTRODUCTION The aim of the INTER project is to investigate fundamental issues related to the design and fabrication of a new generation of microsystems applicable as neural prostheses. A global overview for a peripheral nervous system (PNS) remoted limb prostheses is given in [1]. The scheme configuration is given in figure 1. Nerve signals will be recorded and amplified by a neurosensor which is a regeneration-type sensor. The recorded signals will be processed by an artificial neural net (ANN), which classifies the resulting signals in order to assign certain limb movements. Signals from sensors of the prostheses will be processed by an ANN and transmitted via a signal generat...