Solving optimal stopping problems with deep learning
Taisa Calvette
Rio de Janeiro, State of Rio de Janeiro
- 0 Collaborators
The main objective is to develop a deep learning algorithm for optimal stopping problems, which is a class of stochastic control problems. It is usually used to find the best time to enter (or exit) given the criteria that maximizes the expected gain (or minimize the expected loss). ...learn more
Project status: Concept
oneAPI, Artificial Intelligence
Intel Technologies
DevCloud,
oneAPI,
Intel Python,
OpenVINO
Overview / Usage
The optimal stopping problems are a class of stochastic control problems. It is usually used to find the best time to enter (or exit) given the criteria that maximizes the expected gain (or minimize the expected loss). It is, in general, a nonlinear partial differential equation in the value function and once this solution is known, it can be used to obtain the optimal control time. The main objective of the research is to propose an approach to solve optimal timing problems with the use of deep learning algorithms.
Technologies Used
- TensorFlow .
- Intel's OpenVINO Toolkit for optimization.
- Intel® oneAPI DevCloud for Training.