Systems, Man, and Cybernetics

 

Home

• [P117] Reinforcement Learning for Human-Machine Collaborative Optimization: Application in Ground Water Monitoring
  Author(s): Meghna Babbar-Sebens, Snehasis Mukhopadhyay
  
  

  Abstract
  In this paper, we introduce reinforcement learning as an interactive methodology to solve complex multi-criteria optimization problems for ground water monitoring design. Design decisions are based on multiple criteria that can be obtained quantitatively or interactively via human feedback during the optimization process. For experimental purposes in this paper, human feedback is simulated by adding random noise to a pre-selected quantitative representation of the decision criteria. In practice, the human feedback would be obtained interactively during the optimization process, via human-computer interfaces. Different learning automata based reinforcement learning methods as well as a genetic algorithm based method are used in the experimental studies, which demonstrate the efficiency of the reinforcement learning approaches for interactive optimization.

  
   

Full Text (PDF)