REVIEW 15 cited by
Mirror Descent Policy Optimization
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Mirror Descent Policy Optimization
read the original abstract
Mirror descent (MD), a well-known first-order method in constrained convex optimization, has recently been shown as an important tool to analyze trust-region algorithms in reinforcement learning (RL). However, there remains a considerable gap between such theoretically analyzed algorithms and the ones used in practice. Inspired by this, we propose an efficient RL algorithm, called {\em mirror descent policy optimization} (MDPO). MDPO iteratively updates the policy by {\em approximately} solving a trust-region problem, whose objective function consists of two terms: a linearization of the standard RL objective and a proximity term that restricts two consecutive policies to be close to each other. Each update performs this approximation by taking multiple gradient steps on this objective function. We derive {\em on-policy} and {\em off-policy} variants of MDPO, while emphasizing important design choices motivated by the existing theory of MD in RL. We highlight the connections between on-policy MDPO and two popular trust-region RL algorithms: TRPO and PPO, and show that explicitly enforcing the trust-region constraint is in fact {\em not} a necessity for high performance gains in TRPO. We then show how the popular soft actor-critic (SAC) algorithm can be derived by slight modifications of off-policy MDPO. Overall, MDPO is derived from the MD principles, offers a unified approach to viewing a number of popular RL algorithms, and performs better than or on-par with TRPO, PPO, and SAC in a number of continuous control tasks. Code is available at \url{https://github.com/manantomar/Mirror-Descent-Policy-Optimization}.
Forward citations
Cited by 15 Pith papers
-
On the Policy Convergence of Policy Mirror Descent Methods
Unregularized PMD with any constant step size converges to a limiting optimal policy for general decomposable Legendre mirror maps, with behavior governed by differentiability of ψ at 0 and 1.
-
Priced Motion Through Optimal Faces: A Normal-Fan Geometry for Non-Stationary Adversarial MDPs
Introduces priced face-crossing via normal-fan geometry on occupancy polytopes to decompose dynamic regret into intrinsic motion cost plus within-face error in non-stationary adversarial MDPs.
-
Randomized Advantage Transformation (RAT): Computing Natural Policy Gradients via Direct Backpropagation
RAT reformulates regularized natural policy gradients as vanilla gradients with a transformed advantage, computed efficiently via randomized block Kaczmarz iterations on on-policy data.
-
Post Reasoning: Improving the Performance of Non-Thinking Models at No Cost
Post-Reasoning boosts LLM accuracy by reversing the usual answer-after-reasoning order, delivering mean relative gains of 17.37% across 117 model-benchmark pairs with zero extra cost.
-
Mirror Descent-Ascent for mean-field min-max problems
Establishes O(N^{-1/2}) convergence for simultaneous MDA and O(N^{-2/3}) for alternating MDA to mixed Nash equilibria in mean-field convex-concave min-max problems via dual-space Bregman analysis.
-
Credit Assignment with Resets in Language Model Reasoning
The paper introduces Random-Reset Policy Optimization (RRPO) and Self-Reset Policy Optimization (SRPO) that use resets to enable more precise credit assignment in RL for language model reasoning, with SRPO outperformi...
-
Listwise Policy Optimization: Group-based RLVR as Target-Projection on the LLM Response Simplex
LPO reframes group-based RLVR as explicit target-projection on the LLM response simplex and performs exact divergence minimization to achieve monotonic listwise improvement with bounded gradients.
-
Listwise Policy Optimization: Group-based RLVR as Target-Projection on the LLM Response Simplex
Listwise Policy Optimization explicitly performs target-projection on the LLM response simplex, unifying and improving group-based RLVR methods with monotonic improvement and flexible divergences.
-
Reinforcement Learning with Discrete Diffusion Policies for Combinatorial Action Spaces
A method trains discrete diffusion policies for combinatorial RL by matching to a PMD-regularized target distribution, reporting SOTA performance and sample efficiency on DNA generation, macro-action, and multi-agent ...
-
Dissecting Discrete Soft Actor-Critic: Limitations and Principled Alternatives
Shows entropy coupling limits DSAC on discrete tasks and introduces a generalized actor-critic framework with m-step critics and novel entropy-regularized objectives that perform robustly on Atari.
-
Muon is Scalable for LLM Training
Muon optimizer with weight decay and update scaling achieves ~2x efficiency over AdamW for large LLMs, shown via the Moonlight 3B/16B MoE model trained on 5.7T tokens.
-
A Single Diffusion-Policy Controller for Multi-Task Block Pushing with Zero-Shot Sim-to-Real Transfer
One diffusion policy trained via energy-guided RL solves multi-shape block pushing without demos and transfers zero-shot to real robots under varied conditions.
-
Randomized Advantage Transformation (RAT): Computing Natural Policy Gradients via Direct Backpropagation
RAT estimates Tikhonov-regularized natural policy gradients by rewriting them with the Woodbury identity, approximating the transformed advantage via randomized block Kaczmarz, and applying it as a vanilla policy grad...
-
Optimal Posterior Sampling for Policy Identification in Tabular Markov Decision Processes
A new posterior sampling algorithm for (ε, δ)-PAC policy identification in tabular MDPs achieves asymptotic optimality in sample complexity and posterior contraction rate with O(S²AH) runtime per episode.
-
Stop Overthinking: A Survey on Efficient Reasoning for Large Language Models
A survey organizing techniques to achieve efficient reasoning in LLMs by shortening chain-of-thought outputs.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.