Risk-Sensitivity in Multi-Armed Bandits


  • Empirically surveyed the existing methods for risk-sensitivity in stochastic bandit problems, spanning risk measures such as Variance, Value at Risk (VaR) and conditional Value at Risk (cVaR)
  • Implemented multiple risk-sensitive algorithms for each measure and performed a qualitative and quantitative analysis
  • Introduced novel modifications of the Explore-Then-Commit algorithm for VaR and cVaR measures; both showing performance competent with existing risk-sensitive algorithms