Neural network biometrics - to-self-correcting code converter without redundancy

Classical error detecting and correcting codes have significant redundancy, which makes them ineffective in biometric applications. With redundancy more than 50% the classical codes use their primary resources for correcting errors in redundant parts. The author suggests detecting error syndromes by hashing informational pieces of a code that allows one to abandon self-correcting codes redundancy at all in biometric applications. The article describes code recovery procedures and evaluates self-correction time expenditures