43. Dynamic Programming using Average-Value-at-Risk Criteria
Invited abstract in session MB-11: Optimal and stochastic optimal control 1, stream Optimal and stochastic optimal control.
Monday, 10:30-12:30Room: B100/5017
Authors (first author is the speaker)
| 1. | Kerem Ugurlu
|
| Mathematics, Nazarbayev University |
Abstract
We investigate dynamic programming equations on Averaga-Value-at-Risk (AVaR) using machine learning techniques and demonstrate several simulations. The dynamic programming equation on AVaR is specifically using a so called "state aggregation technique" that makes use of the sufficient statistic of the optimization problem. Via state aggregation, we present the "Markovian" framework and demonstrate several implementations using fully connected neural networks. The risk averseness level alpha on the model is also investigated via several simulations.
Keywords
- Optimization under uncertainty
- AI based optimization methods
- Distributionally robust optimization
Status: accepted
Back to the list of papers