Deterministic Uncertainty Propagation for Improved Model-Based Offline Reinforcement Learning

• URL: http://arxiv.org/abs/2406.04088v1
• Date: Thu, 6 Jun 2024 13:58:41 GMT
• Title: Deterministic Uncertainty Propagation for Improved Model-Based Offline Reinforcement Learning
• Authors: Abdullah Akgül, Manuel Haußmann, Melih Kandemir,
• Abstract summary: Current approaches to model-based offline Reinforcement Learning (RL) often incorporate uncertainty-based reward penalization. We argue that this penalization introduces excessive conservatism, potentially resulting in suboptimal policies through underestimation. We identify as an important cause of over-penalization the lack of a reliable uncertainty estimator capable of propagating uncertainties in the Bellman operator.
• Score: 12.490614705930676
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Current approaches to model-based offline Reinforcement Learning (RL) often incorporate uncertainty-based reward penalization to address the distributional shift problem. While these approaches have achieved some success, we argue that this penalization introduces excessive conservatism, potentially resulting in suboptimal policies through underestimation. We identify as an important cause of over-penalization the lack of a reliable uncertainty estimator capable of propagating uncertainties in the Bellman operator. The common approach to calculating the penalty term relies on sampling-based uncertainty estimation, resulting in high variance. To address this challenge, we propose a novel method termed Moment Matching Offline Model-Based Policy Optimization (MOMBO). MOMBO learns a Q-function using moment matching, which allows us to deterministically propagate uncertainties through the Q-function. We evaluate MOMBO's performance across various environments and demonstrate empirically that MOMBO is a more stable and sample-efficient approach.

Related papers

• Sequential Monte Carlo for Policy Optimization in Continuous POMDPs [9.690099639375456]
  We introduce a novel policy optimization framework for continuous partially observable Markov decision processes (POMDPs)<n>Our method casts policy learning as probabilistic inference in a non-Markovian Feynman--Kac model.<n>We demonstrate the effectiveness of our algorithm across standard continuous POMDP benchmarks.
  arXiv  Detail & Related papers  (2025-05-22T14:45:46Z)
• Bayesian learning of the optimal action-value function in a Markov decision process [7.186805722297615]
  We provide a full Bayesian framework, from modelling to inference to decision-making.<n>For inference, we propose an adaptive sequential Monte Carlo algorithm to both sample from and adjust the sequence of relaxed posterior distributions.<n>While commonly done, we provide new insight that clearly shows that it is a generalisation of Thompson sampling from multi-arm bandit problems.
  arXiv  Detail & Related papers  (2025-05-03T16:37:14Z)
• To bootstrap or to rollout? An optimal and adaptive interpolation [4.755935781862859]
  We introduce a class of Bellman operators that interpolate between bootstrapping and rollout methods. Our estimator combines the strengths of the bootstrapping-based temporal difference (TD) estimator and the rollout-based Monte Carlo (MC) methods.
  arXiv  Detail & Related papers  (2024-11-14T19:00:00Z)
• Self-Evolutionary Large Language Models through Uncertainty-Enhanced Preference Optimization [9.618391485742968]
  Iterative preference optimization has recently become one of the de-facto training paradigms for large language models (LLMs) We present an uncertainty-enhanced textbfPreference textbfOptimization framework to make the LLM self-evolve with reliable feedback. Our framework substantially alleviates the noisy problem and improves the performance of iterative preference optimization.
  arXiv  Detail & Related papers  (2024-09-17T14:05:58Z)
• Offline Bayesian Aleatoric and Epistemic Uncertainty Quantification and Posterior Value Optimisation in Finite-State MDPs [3.1139806580181006]
  We address the challenge of quantifying Bayesian uncertainty in offline use cases of finite-state Markov Decision Processes (MDPs) with unknown dynamics. We use standard Bayesian reinforcement learning methods to capture the posterior uncertainty in MDP parameters. We then analytically compute the first two moments of the return distribution across posterior samples and apply the law of total variance. We highlight the real-world impact and computational scalability of our method by applying it to the AI Clinician problem.
  arXiv  Detail & Related papers  (2024-06-04T16:21:14Z)
• Uncertainty Quantification for Traffic Forecasting: A Unified Approach [21.556559649467328]
  Uncertainty is an essential consideration for time series forecasting tasks. In this work, we focus on quantifying the uncertainty of traffic forecasting. We develop Deep S-Temporal Uncertainty Quantification (STUQ), which can estimate both aleatoric and relational uncertainty.
  arXiv  Detail & Related papers  (2022-08-11T15:21:53Z)
• Regularization Guarantees Generalization in Bayesian Reinforcement Learning through Algorithmic Stability [48.62272919754204]
  We study generalization in Bayesian RL under the probably approximately correct (PAC) framework. Our main contribution is showing that by adding regularization, the optimal policy becomes stable in an appropriate sense.
  arXiv  Detail & Related papers  (2021-09-24T07:48:34Z)
• COMBO: Conservative Offline Model-Based Policy Optimization [120.55713363569845]
  Uncertainty estimation with complex models, such as deep neural networks, can be difficult and unreliable. We develop a new model-based offline RL algorithm, COMBO, that regularizes the value function on out-of-support state-actions. We find that COMBO consistently performs as well or better as compared to prior offline model-free and model-based methods.
  arXiv  Detail & Related papers  (2021-02-16T18:50:32Z)
• Amortized Conditional Normalized Maximum Likelihood: Reliable Out of Distribution Uncertainty Estimation [99.92568326314667]
  We propose the amortized conditional normalized maximum likelihood (ACNML) method as a scalable general-purpose approach for uncertainty estimation. Our algorithm builds on the conditional normalized maximum likelihood (CNML) coding scheme, which has minimax optimal properties according to the minimum description length principle. We demonstrate that ACNML compares favorably to a number of prior techniques for uncertainty estimation in terms of calibration on out-of-distribution inputs.
  arXiv  Detail & Related papers  (2020-11-05T08:04:34Z)
• Control as Hybrid Inference [62.997667081978825]
  We present an implementation of CHI which naturally mediates the balance between iterative and amortised inference. We verify the scalability of our algorithm on a continuous control benchmark, demonstrating that it outperforms strong model-free and model-based baselines.
  arXiv  Detail & Related papers  (2020-07-11T19:44:09Z)
• A Distributional Analysis of Sampling-Based Reinforcement Learning Algorithms [67.67377846416106]
  We present a distributional approach to theoretical analyses of reinforcement learning algorithms for constant step-sizes. We show that value-based methods such as TD($lambda$) and $Q$-Learning have update rules which are contractive in the space of distributions of functions.
  arXiv  Detail & Related papers  (2020-03-27T05:13:29Z)
• Distributionally Robust Bayesian Quadrature Optimization [60.383252534861136]
  We study BQO under distributional uncertainty in which the underlying probability distribution is unknown except for a limited set of its i.i.d. samples. A standard BQO approach maximizes the Monte Carlo estimate of the true expected objective given the fixed sample set. We propose a novel posterior sampling based algorithm, namely distributionally robust BQO (DRBQO) for this purpose.
  arXiv  Detail & Related papers  (2020-01-19T12:00:33Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.