We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.LG

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Accelerated Target Updates for Q-learning

Abstract: This paper studies accelerations in Q-learning algorithms. We propose an accelerated target update scheme by incorporating the historical iterates of Q functions. The idea is conceptually inspired by the momentum-based accelerated methods in the optimization theory. Conditions under which the proposed accelerated algorithms converge are established. The algorithms are validated using commonly adopted testing problems in reinforcement learning, including the FrozenLake grid world game, two discrete-time LQR problems from the Deepmind Control Suite, and the Atari 2600 games. Simulation results show that the proposed accelerated algorithms can improve the convergence performance compared with the vanilla Q-learning algorithm.
Comments: We need further adjustment of some parts of the papaer
Subjects: Machine Learning (cs.LG); Optimization and Control (math.OC); Machine Learning (stat.ML)
Cite as: arXiv:1905.02841 [cs.LG]
  (or arXiv:1905.02841v2 [cs.LG] for this version)

Submission history

From: Huaqing Xiong [view email]
[v1] Tue, 7 May 2019 23:14:23 GMT (4164kb,D)
[v2] Sat, 11 May 2019 12:40:11 GMT (0kb,I)

Link back to: arXiv, form interface, contact.