Real-Time Dynamic Programming (RTDP) Algorithm

A Real-Time Dynamic Programming (RTDP) Algorithm is a Dynamic Programming Algorithm that is enhanced by implementing a heuristic search algorithm.

• Context:
  • It is an Adaptive Real-Time Dynamic Programming (ARTDP) Algorithm without a system identification.
• Example(s):
  • Bayesian Real-Time Dynamic Programming (BRTDP) Algorithm (Sanner et al., 2009),
  • Bounded Real-Time Dynamic Programming (BRTDP) Algorithm (McMahan et al., 2005),
  • Labeled Real-Time Dynamic Programming (LRTDP) Algorithm (Bonet & Geffner, 2003),
  • …
• Counter-Example(s):
  • Brown-UMBC Reinforcement Learning and Planning (BURLAP) Algorithm],
  • Forward-Backward Algorithm.
• See: Anytime Algorithm; Approximate Dynamic Programming; Reinforcement Learning.

---

---

References

2017

• (Sammut & Webb, 2017) ⇒ (2017) Real-Time Dynamic Programming. In: Sammut, C., Webb, G.I. (eds) Encyclopedia of Machine Learning and Data Mining. Springer, Boston, MA.
  • QUOTE: Real-Time Dynamic Programming (RTDP) is the same as Adaptive Real-Time Dynamic Programming (ARTDP) without the system identification component(...)

    (...)RTDP combines strengths of heuristic search and DP. Like heuristic search – and unlike conventional DP – it does not have to evaluate the entire state space in order to produce an optimal solution. Like DP – and unlike most heuristic search algorithms – it is applicable to nondeterministic problems. Additionally, RTDP's performance as an anytime algorithm is better than conventional DP and heuristic search algorithms. ARTDP extends these strengths to problems for which a good model is not initially available.

2009

• (Sanner et al., 2009) ⇒ Scott Sanner, Robby Goetschalckx, Kurt Driessens, and Guy Shani (2009). "Bayesian Real-Time Dynamic Programming". In: In Proceedings of the 21st International Joint Conference on Artificial Intelligence (IJCAI-09).

2005

• (McMahan et al., 2005) ⇒ H. Brendan McMahan, Maxim Likhachev, and Geoffrey J. Gordon (2005). Bounded Real-Time Dynamic Programming: RTDP with monotone upper bounds and performance guarantees. In: In Proceedings of the 22nd International Conference on Machine learning (pp. 569-576).

2003

• (Bonet & Geffner, 2003) ⇒ Blai Bonet, and Hector Geffner (2003). "Labeled RTDP: Improving the Convergence of Real-Time Dynamic Programming". In: Proceedings of Thirteenth Interational Conference on Automated Planning and Scheduling (ICAPS), Vol. 3, pp. 12-21.

---