• A
  • A
  • A
  • ABC
  • ABC
  • ABC
  • А
  • А
  • А
  • А
  • А
Regular version of the site
Menu
  • HSE University
  • Publications of HSE
  • Articles
  • Возможности методов искусственного интеллекта для выявления и использования новых знаний на примере задачи управления персоналом

Article

Возможности методов искусственного интеллекта для выявления и использования новых знаний на примере задачи управления персоналом

Ясницкий Л. Н., Михалева Ю. А., Черепанов Ф. М.

Developed a computer program designed to assess psychological capacity of chief of staff (CPD), which refers to a person's ability to perform leadership. The program is based on neural network, trained on the results of the survey of employees of a number of large organizations. Mathematical model analysis revealed several unexpected patterns. It turned out that CPD, in addition to the traditional parameters (age, gender, type of activity, number of children, and so on), depends also on factors to explain the effect of which in the framework of existing scientific knowledge is not possible. Factors that typically use astrologers in the preparation of horoscopes. Acceptable accuracy of neural network mathematical model was able to build using as input parameters at the same time factors such as astrological and australopitecus nature. Attempts to exclude from the number of input parameters of neural network mathematical model of any group of these two groups of factors have led to a sharp increase in the error model. Research on the impact of changes in input parameters on the simulation result (CPD) performed by "freezing" (commit) some of the input parameters and gradual changes in other input parameters while monitoring the output signal value of the network. These studies helped to identify the dependence of the ability of people to the leadership of their gender, age, number of children, type of activity, the sign of the Zodiac. Constructed a histogram of the distribution of the value of all input parameters. It is noted that the use of unexplained factors as input parameters of the mathematical models is justified by the fact that in some cases it can significantly improve the accuracy of the method of mathematical modeling and its capabilities. It is also noted that neural network technology developed in the work, are a useful tool for the identification and use of new, previously unknown, and as yet unexplained regularities.