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.

168,932 papers · 148 categories

Trend · papers per month

count % of corpus

7 results for “trajectory-wise”

Trajectory-wise Control Variates for Variance Reduction in Policy Gradient Methodscs.LG

Trajectory-wise CVs reduce variance in policy gradient methods.

problem High variance in estimating policy gradient estimates.

method Proposes trajectory-wise control variates to reduce variance without bias.

result Trajectory-wise CVs are optimal for variance reduction under reasonable assumptions.

policy gradient variance reduction control variates reinforcement learning trajectory-wise

2019-08-08abs ↗pdf ↗

Provably Efficient Offline Reinforcement Learning with Trajectory-Wise Rewardcs.LG

A new algorithm for offline RL with trajectory-wise reward reduces bias and variance errors.

problem Offline RL with trajectory-wise reward incurs large bias and variance errors.

method PARTED algorithm that decomposes trajectory return into proxy rewards and performs pessimistic value iteration.

result PARTED achieves provably efficient suboptimality bounds in general MDPs with trajectory-wise reward.

offline reinforcement learning trajectory-wise reward value iteration suboptimality bounds reinforcement learning algorithms

2022-06-13abs ↗pdf ↗

Differentially Private Reinforcement Learning with Self-Playcs.LG

Study differential privacy in multi-agent RL, achieving efficient and private learning.

problem Protecting sensitive data in multi-agent reinforcement learning.

method Extending DP definitions to two-player games, designing an efficient algorithm with privatized bonuses.

result Achieved trajectory-wise differential privacy in multi-agent RL, improving regret bounds.

Differential Privacy Reinforcement Learning Multi-Agent Systems Privacy Protection

2024-04-11abs ↗pdf ↗

Efficient and Unbiased Sampling from Boltzmann Distributions via Variance-Tuned Diffusion Modelscs.LG

VT-DIS improves sampling from Boltzmann distributions with minimal overhead.

problem Bias in Monte Carlo estimates from score-based diffusion models.

method Variance-Tuned Diffusion Importance Sampling (VT-DIS) adapts noise covariance to correct bias.

result VT-DIS achieves effective sample sizes of 80%, 35%, and 3.5% on benchmarks, using less computational budget.

score-based diffusion models Monte Carlo sampling Boltzmann distributions Variance-tuning Importance sampling

2025-05-27abs ↗pdf ↗

A bandit-learning approach to multifidelity approximationmath.NA

Paper introduces a bandit-learning method for multifidelity approximations.

problem Efficiently using data of varying fidelities in scientific computation.

method Formulates multifidelity approximation as a modified stochastic bandit problem and proposes AETC algorithm.

result Established optimality of AETC algorithm for multifidelity approximation.

bandit learning multifidelity approximation stochastic bandit scientific computation parameter estimation

2021-03-29abs ↗pdf ↗

The roughness exponent and its model-free estimationmath.ST

Estimates roughness of stochastic processes without assuming specific models.

problem Estimating roughness of stochastic processes without assuming specific models.

method Using Faber-Schauder coefficients and martingales, we provide a method to estimate the roughness exponent of stochastic processes.

result The roughness exponent can be estimated without assuming specific models, providing a strong consistency result for the Gladyshev estimators.

roughness exponent stochastic processes Faber-Schauder coefficients Gladyshev estimators martingales

2021-11-19abs ↗pdf ↗

Reinforcement Learning from Human Feedback without Reward Inference: Model-Free Algorithm and Instance-Dependent Analysiscs.LG

New RLHF algorithm identifies optimal policies from human feedback without explicit reward inference.

problem Training large language models with human feedback without reward inference.

method Model-free RLHF algorithm

\mathsf{BSAD}

that identifies optimal policies directly from human preference.

result Provable, instance-dependent sample complexity

ilde{\mathcal{O}}(c_{\mathcal{M}}SA^3H^3M\log\frac{1}δ)

Reinforcement Learning Human Feedback Model-Free Reward Inference Best Policy Identification

2024-06-11abs ↗pdf ↗