The paper presents an intelligent system for entry control. It consists of four intelligent modules. Each of them classifies an entry event as normal event or potentially dangerous event. The integrated classification increases entry security in a building and with some modifications also in other internet systems.
COBISS.SI-ID: 22561319
In this paper we present a high-security access control system using traditional biometrical sensors and several intelligent methods for access control, which intend to prevent unauthorized entries even if the sensors are by-passed. The system enables integration of different sensors and AI modules. The new developed modules learn to recognize and distinguish between regular and irregular attempts of access. The combination of learning methods with classification trees and k-NN significantly improves performance as shown in the tests.
COBISS.SI-ID: 21347879