# Offline Deep Q* Estimation with Diffusion Models

arXiv: https://arxiv.org/abs/2608.14401 · HTML: https://mlsanity.com/paper/2608.14401/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Diffusion and flow policies for robotics, one of 14 directions in Diffusion models.
- Builds on 0 of its 101 references that are in the map.
- First posted on arXiv 14 Aug 2026.

## Placed in

- Diffusion models / [Diffusion and flow policies for robotics](https://mlsanity.com/diffusion/diffusion-and-flow-policies-for-robotics/)
- Offline RL and imitation learning from data / [Offline RL algorithms](https://mlsanity.com/offlinerl/offline-rl-algorithms/)

## Abstract (arXiv)

In offline RL, estimating the optimal action-value function Q^* can be formulated as solving the optimal Bellman equation based solely on offline observations. A fundamental challenge is that the reward function and transition kernel are unknown, so the optimal Bellman operator is not directly observable from data. To address this issue, we propose a novel framework that decouples operator estimation from value function learning. In this approach, we first formulate conditional diffusion models to estimate the reward law and transition kernel, which induces a data-driven approximation of the optimal Bellman operator. We then plug these estimators into the Bellman equation and obtain a deep estimator of Q^* by minimizing the empirical Bellman residual over a neural network function class. Theoretically, we first establish sharp nonasymptotic convergence rates for learning the optimal Bellman operator through an end-to-end analysis of conditional diffusion estimation in total variation distance. We then establish the oracle value-stage rate widetildemathcal Obigl(n^-frac2βdx+da+2βbigr) for the excess Bellman residual risk. Finally, under a concentrability condition, we translate this residual bound into an L2 convergence rate of widetildemathcal Obigl(n^-fracβdx+da+2βbigr) for the resulting deep estimator of Q^*, where dx and da denote the dimensions of the state and action spaces, respectively, and β denotes the H"older smoothness index of Q^*. Importantly, our theoretical analysis does not rely on completeness assumptions commonly used in deep RL theory. Extensive numerical experiments demonstrate the effectiveness of the proposed method and its strong empirical performance.