The Digital Twin of an Organization by Utilizing Reinforcing Deep Learning

Research output: Chapter in Book/Report/Conference proceedingChapterScientificpeer-review


Chapter deals with latest knowledge on deep reinforcement learning in the context of organizational management. Article presents reinforcement learning (RL) as a tool for the manager on the path to learning winning behavior in the complex environment of organization management. Organization management has wicked learning challenges because agents are under biases that prevent understanding the phenomenon of delayed reward. Therefore, the digital simulation with RL is effective forming breakthrough learning results. Human capital management theories provide architecture in creating organization digital twin where agent can practice management actions effect on business economics and staff wellbeing. Utilizing RL algorithms, it is possible to foster behavior for creating sustainable competitive advantage – this means the Nash equilibrium between profit and staff wellbeing. In this digital twin there is AI learning assistant as a teacher that provides demonstrations on how to act so that the delayed reward is good in the future. The article explains game theoretical approach that is the foundation for creating management deep learning AI system. Human agent at the organization is playing the game of Strategic Stochastic Bayesian Nonsymmetric Signaling game in co-operative or non-cooperative way and at zero-sum or general sum game mind-set.
Original languageEnglish
Title of host publicationDeep learning applications
EditorsPier Luigi Mazzeo, Paolo Spagnolo
Place of PublicationLondon
ISBN (Electronic)9781839623752, 9781839623769
Publication statusPublished - 11 Feb 2021
MoEC publication typeA3 Part of a book or another research book


  • Reinforcement learning
  • Digital twin
  • QWL
  • Management
  • Game theory

Field of science

  • Administrative science


Dive into the research topics of 'The Digital Twin of an Organization by Utilizing Reinforcing Deep Learning'. Together they form a unique fingerprint.

Citation for this output