Call for papers
We will invite submissions on topics such as, but not limited to:
- Optimization formulations and algorithms for RL
- Optimization correspondences of the components in RL (e.g., on/off-policy, exploration, replay buffer, entropy regularization, target networks, reward shaping, variance reduction, stability, and generalization, etc)
- Theoretical analysis of existing RL algorithms (e.g., temporal difference algorithms);
- Optimization for other RL settings (e.g., distributed/multi-agent RL, robust RL, partial observed RL, and hierarchical RL)
- Empirical comparison of optimization algorithms in RL applications and benchmarks Other topics at the intersection between reinforcement learning and optimization
- Open problems in optimization raised in RL tasks
A submission is up to 6 pages long in the NeurIPS style, excluding references and appendices. The submission process will be handled via CMT. Author names need not be anonymised. Parallel submissions (e.g., AISTATS and ICLR) are permitted.
The submission deadline is September 10th September 17th, 2019, 11:59pm AOE and the acceptance notification will be distributed no later than October 1st, 2019. Submissions will be accepted as contributed talks, spotlight or poster presentations based on novelty, technical merit and alignment to the workshop’s goals. Final versions will be posted on the workshop website.
Paper submission portal: https://cmt3.research.microsoft.com/OPTRL2019
Camera-ready papers
A submission is up to 6 pages long in the OptRL style, excluding references and appendices.
For camera-ready version, please use the “final” option:
\usepackage[final]{optrl_2019} %produce camera-ready copy
The submission site is https://cmt3.research.microsoft.com/OPTRL2019 . The final versions will be posted on the workshop website.
Poster size
Posters are taped onto the wall. Posters should be on lightweight paper, not laminated one. We will provide the tape. The posters should be no larger than 36W x 48H inches or 90 x 122 cm in portrait mode.