Lune

Abstract

In this paper, we present a policy gradient method that avoids exploratory noise injection and performs policy search over the deterministic landscape. By avoiding noise injection all sources of estimation variance can be eliminated in systems with deterministic dynamics (up to the initial state distribution). Since deterministic policy regularization is impossible using traditional non-metric measures such as the KL divergence, we derive a Wasserstein-based quadratic model for our purposes. We state conditions on the system model under which it is possible to establish a monotonic policy improvement guarantee, propose a surrogate function for policy gradient estimation, and show that it is possible to compute exact advantage estimates if both the state transition model and the policy are deterministic. Finally, we describe two novel robotic control environments-one with non-local rewards in the frequency domain and the other with a long horizon (8000 timesteps)-for which our policy gradient method (TDPO) significantly outperforms existing methods (PPO, TRPO, DDPG, and TD3). Our implementation with all the experimental settings is available at https://github.com/ehsansaleh/code_tdpo .

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

3 top-tier papers have cited it since. Know what they found before your reviewers do.

Questions to start from

Lune can draw on

the full paper · the top-tier papers that cite it · related work

Get the answer

Free to start. No credit card required.

Lineage

  • Convergence Analysis of Policy Gradient Methods with Dynamic Stochasticity cites Learning Optimal Deterministic Policies with Stochastic Policy Gradients
Builds onCited byLeft to right by year; the more cited, the higher.

Cited by top-tier papers3

  1. Learning Optimal Deterministic Policies with Stochastic Policy GradientsAlessandro Montenegro, Marco Mussi, Alberto Maria Metelli, Matteo PapiniICML 2024 · 11 citations
  2. Augmented Bayesian Policy SearchMahdi Kallel, Debabrota Basu, Riad Akrour, Carlo D'EramoICLR 2024 · 4 citations
  3. Convergence Analysis of Policy Gradient Methods with Dynamic StochasticityAlessandro Montenegro, Marco Mussi, Matteo Papini, Alberto Maria MetelliICML 2025

Builds on2

  1. Statistically Efficient Off-Policy Policy GradientsNathan Kallus, Masatoshi UeharaICML 2020 · 43 citations
  2. Learning to Score Behaviors for Guided Policy OptimizationAldo Pacchiano, Jack Parker-Holder, Yunhao Tang, Krzysztof Choromanski et al.ICML 2020 · 42 citations

Related papers

  1. Deterministic Policy Gradient Primal-Dual Methods for Continuous-Space Constrained MDPsSergio Rozada, Dongsheng Ding, Antonio G. Marques, Alejandro RibeiroAAAI 2025 · 2 citations
  2. Policy Search by Target Distribution Learning for Continuous ControlChuheng Zhang, Yuanqi Li, Jian LiAAAI 2020 · 6 citations
  3. Adaptive-Gradient Policy Optimization: Enhancing Policy Learning in Non-Smooth Differentiable SimulationsFeng Gao, Liangzhi Shi, Shenao Zhang, Zhaoran Wang et al.ICML 2024 · 7 citations
  4. Trust Region Policy Optimization with Optimal Transport Discrepancies: Duality and Algorithm for Continuous ActionsAntonio Terpin, Nicolas Lanzetti, Batuhan Yardim, Florian Dörfler et al.NeurIPS 2022 · 14 citations
  5. Absolute Policy Optimization: Enhancing Lower Probability Bound of Performance with High ConfidenceWeiye Zhao, Feihan Li, Yifan Sun, Rui Chen et al.ICML 2024 · 5 citations
Dusk over the sea between two cliffs drawn in fine vertical lines