Influence of the signer's psychophysiological state on the results of his identification using handwritten pattern by natural and artificial intelligence

Abstract

At present, while various mechanisms to ensure information security are actively being improved, particular attention is paid to prevent unauthorized access to information resources. The human factor and process of identification still remain the most problematic, as well as user authentication. A progress in the technology of information resources protection from internal security threats paves its way towards biometric systems of hidden identification of computer users and their psychophysiological state. A change in psychophysiological state results in the person's handwriting. The influence of the signer’s state of fatigue and excitation on the results of its identification both by a person and by pattern recognition methods on reproduced signatures are studied. Capabilities of human and artificial intelligence are compared in equal conditions. When the state of the signer changes, the probability of erroneous recognition by artificial intelligence increases by factor 3.3 to 3.7. A person identifies a handwritten image with fewer errors in case when the signer is agitated, and with higher error rate if the signer is tired.

Keywords

• identification features
• appearance of the signature
• password handwriting dynamics
• pattern recognition algorithm by human and machine
• Bayes' formula