Research
On-device research index

arXiv research

A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.

168,694 papers · 148 categories

Trend · papers per month

57114171228 · Jun 202019922001200920172026
48 results for policy diffusion

This work extends ME-RL using diffusion models to sample optimal policies.

problem Sampling from the optimal policy trajectory distribution in ME-RL.
method Introducing Diffusion-Augmented Markov Decision Processes (DA-MDPs) to minimize reverse KL divergence.
result DA-MDPs enable seamless integration into various ME-RL methods and outperform baselines.

Diffusion-QL uses diffusion models to improve offline RL performance.

problem Offline RL struggles with function approximation errors on out-of-distribution actions.
method Diffusion-QL represents the policy as a conditional diffusion model and optimizes action-values.
result Diffusion-QL achieves state-of-the-art performance on D4RL benchmark tasks.

Improved exploration in cooperative multi-agent reinforcement learning.

problem Limited expressiveness of Gaussian policies in DecSPG hinders effective exploration.
method Proposes decentralized diffusion policy learning (DDPL) with denoising diffusion probabilistic models.
result Consistently improved performance on various MARL benchmarks.

This paper identifies drift Lipschitz budget K as key to diffusion policy expressivity and statistical trade-offs.

problem Understanding and maximizing the expressivity of diffusion policies while managing statistical limitations.
method Identifying drift Lipschitz budget K as central, quantifying expressivity and statistical behavior, proving lower bounds, and providing practical implementation guidelines.
result Balancing expressivity and statistical complexity yields a finite-sample performance gap, with rates depending on sample size and drift type.

Study on policy gradient for stochastic bandits using diffusion approximation.

problem Improving policy gradient methods for stochastic bandits with optimal regret bounds.
method Continuous-time diffusion approximation of policy gradient with learning rate analysis.
result Proved optimal regret bound of O(klog(k)log(n)/η)O(k \log(k) \log(n) / η) for η=O(Δ2/log(n))η= O(Δ^2/\log(n)).

wd1 improves reasoning in dLLMs by optimizing policies without policy ratios.

problem Improving reasoning in diffusion-based large language models through RL.
method wd1: ratio-free policy optimization using weighted log-likelihood.
result wd1 outperforms diffusion-based GRPO while requiring lower computational cost.

CODA resolves coordination issues in offline multi-agent reinforcement learning.

problem Coordination failure in offline multi-agent reinforcement learning.
method Diffusion-based multi-agent trajectory generator for data augmentation.
result CODA resolves coordination pathologies in continuous polynomial games and complex benchmarks.

SDM Policy accelerates inference for robotic tasks while maintaining high action quality.

problem Prolonged inference times in diffusion-based policies for high-frequency control tasks.
method Two-stage optimization: score matching and distribution matching; dual-teacher mechanism.
result 6x inference speedup with state-of-the-art action quality.

Discrete diffusion samplers improve sampling from unnormalised densities.

problem Sampling from discrete unnormalised densities efficiently.
method Introduce off-policy training techniques and data-to-energy Schrödinger bridge training for discrete diffusion samplers.
result Improved performance on synthetic and new benchmarks.

Unified q-learning for mean-field jump-diffusion models with unobservable population distribution.

problem Continuous-time q-learning in mean-field jump-diffusion models with unobservable population distribution.
method Proposed decoupled Iq-function for unified policy evaluation in MFG and MFC problems; unified q-learning algorithm based on test policies and averaged martingale orthogonality condition.
result Unified policy evaluation rule for MFG and MFC problems based on decoupled Iq-function.

RL for jump-diffusions applies to financial portfolio selection and option hedging.

problem Optimizing control in systems with jump-diffusion dynamics.
method Entropy-regularized exploratory control with stochastic policies, using existing diffusion algorithms with modifications.
result RL algorithms and parameterizations are invariant to jumps in jump-diffusion systems.

QAM uses adjoint matching to optimize continuous-action RL policies efficiently.

problem Efficient optimization of expressive diffusion or flow-matching policies with respect to a Q-function.
method QAM leverages adjoint matching to bypass the numerical instability of backpropagation through multi-step denoising processes.
result QAM consistently outperforms prior approaches on hard, sparse reward tasks in offline and offline-to-online RL.

Develops a framework for multi-objective learning in diffusion models with limited labeled data.

problem Achieving good trade-offs in multi-objective learning with diffusion models requires a generalist model class with larger capacity than individual tasks.
method Proposes a two-stage training procedure: first fitting specialist models from limited paired data, then distilling them into a generalist model.
result Establishes generalization bounds showing the number of paired samples depends only on specialist model complexity.

New text-to-image diffusion models improve scene understanding for AI agents.

problem Fine-grained scene understanding for AI agents from text and images.
method Pre-trained text-to-image diffusion models optimized for generating images from text prompts.
result Policies learned with Stable Control Representations outperform state-of-the-art approaches on various control tasks.

Proposes KStar Diffuser for kinematics-aware bimanual robotic manipulation.

problem Challenges in applying imitation learning to bimanual robotic tasks.
method Integrates physical robot structure into action prediction using a dynamic spatial-temporal graph and differentiable kinematics.
result Effective generation of kinematics-aware actions in both simulation and real-world environments.

Study improves sampling efficiency of diffusion models using RL and PDEs.

problem Training neural stochastic differential equations without access to target samples.
method Proves equivalences between RL methods and PDEs, uses coarse time discretization.
result Improves sample efficiency and reduces computational cost.

A new method for offline RL using diffusion models and self-weighted guidance.

problem Challenges in computing scores for offline RL using weight functions.
method Constructing a diffusion over actions and weights, using the diffusion model for guidance.
result Performs on par with state-of-the-art methods on challenging environments.

New RL algorithm GDPO improves DLM reasoning efficiency.

problem Adapting RL to DLMs for efficient, unbiased likelihood estimation.
method Group Diffusion Policy Optimization (GDPO) using semi-deterministic Monte Carlo.
result GDPO outperforms existing methods on math, reasoning, and coding benchmarks.

New method reduces over-pessimism in Bayesian control under parameter uncertainty.

problem Over-pessimism in Bayesian control due to misspecified priors.
method Distributionally robust Bayesian control (DRBC) with strong duality and optimization.
result Validated algorithm on synthetic and real data, reducing over-pessimism.

Study optimal stopping for diffusion processes with unknown primitives, applying RL and martingale methods.

problem Optimal stopping for diffusion processes with unknown model primitives.
method Continuous-time reinforcement learning framework, variational inequality formulation, stochastic optimal control, entropy regularizer, semi-analytical optimal Bernoulli distribution, policy improvement theorem, policy iterations.
result Demonstrated high accuracy in learning value functions and characterizing free boundaries for various optimal stopping problems.

Study how untrained policies explore in RL environments.

problem Challenges in reinforcement learning, especially sparse or adversarial reward structures.
method Theoretical and empirical analysis of untrained deep neural policies in a toy model.
result Untrained policies generate correlated actions and non-trivial state-visitation distributions.

A new framework for generative modeling using value-driven transport.

problem Developing efficient methods for generative modeling.
method A discrete-time stochastic control formulation of measure transport, formulated as a linear program with dual variables corresponding to the optimal value function.
result Well-trained VDT policies lead to straight transport paths that can be simulated quickly and robustly.

New method uses Hermite polynomials for American option valuation.

problem Valuation of American options with complex jump-diffusion dynamics.
method Hermite polynomial expansions of transition density and early exercise premium.
result Converging approximations to true option prices and exercise boundaries.

New framework for policy gradient methods in continuous time reinforcement learning.

problem Addressing policy gradient methods for continuous time reinforcement learning.
method Control randomisation technique to derive policy gradient representation for various Markovian control problems.
result Demonstrated application to optimal switching problems in the energy sector.

New methods improve memory efficiency for sampling from complex distributions.

problem Sampling from complex unnormalized distributions over discrete domains.
method Two novel training methods for discrete diffusion samplers.
result Achieve state-of-the-art results in unsupervised combinatorial optimization.

New method simulates sticky boundaries in multidimensional diffusions.

problem Simulating sticky boundaries in multidimensional diffusions.
method Approximate sticky diffusion by a Markov chain, using either finite difference or matching local moments.
result Validates both construction methods for first-order simulation schemes.

Improved text-to-image and multimodal understanding through adaptive generation order optimization.

problem Determining optimal generation sequences in text-to-image synthesis and multimodal understanding.
method Introduced a learnable control module trained via Group Relative Policy Optimization (GRPO) to determine the generation order.
result Learning the control block substantially improves text-to-image alignment and multimodal understanding in DLMs.

Paper solves dynamic portfolio selection using generative models.

problem Dynamic mean-variance portfolio selection problem in a model-free manner.
method Adaptive training and sampling methods for diffusion models, quantification bounds using adapted Wasserstein metric.
result Proposes a policy gradient algorithm that outperforms baselines on real data.

A RL-based method adds conditional controls to pre-trained diffusion models.

problem Precise control over generated samples in diffusion models.
method Formulates the task as an RL problem, using a classifier and KL divergence as reward functions.
result Produces soft-optimal policies that maximize reward functions, enabling conditional sampling.

Proposes a new RL method to fine-tune flow-based models with arbitrary rewards.

problem Challenges in fine-tuning continuous flow-based generative models with arbitrary reward functions.
method Online Reward-Weighted Conditional Flow Matching with Wasserstein-2 Regularization (ORW-CFM-W2)
result Achieves optimal policy convergence with controllable trade-offs between reward maximization and diversity preservation.

Study indifference pricing for insurance policies in a regime-switching market model.

problem Indifference pricing of pure endowment policies in a stochastic-factor model with different economic regimes.
method Stochastic control approach based on Hamilton-Jacobi-Bellman equation, Feynman-Kac formula, and sensitivity analysis.
result Characterization of indifference price as a solution to a linear PDE and a backward PDE.

Improved RL training for DMs reduces mode collapse and preserves diversity.

problem Mode collapse and training instability in RL fine-tuned diffusion models.
method Dynamic hierarchical RL training with sliding-window parameter regularisation.
result Models trained with HRF achieve better preservation of diversity in downstream tasks.

JADAI optimizes design and inference for parameter estimation.

problem Parameter estimation with active optimization of design variables.
method Jointly trains a policy, history network, and inference network to minimize posterior error.
result Achieves superior or competitive performance across benchmarks.

Tutorial on optimizing diffusion model samples for specific metrics.

problem Optimizing diffusion model samples for specific downstream metrics.
method Review and exploration of inference-time guidance and alignment methods.
result Unified perspective on inference-time algorithms and novel methods.