Safe Learning Under Irreversible Dynamics via Asking for Help

• URL: http://arxiv.org/abs/2502.14043v2
• Date: Tue, 16 Sep 2025 13:55:47 GMT
• Title: Safe Learning Under Irreversible Dynamics via Asking for Help
• Authors: Benjamin Plaut, Juan Liévano-Karim, Hanlin Zhu, Stuart Russell,
• Abstract summary: Most learning algorithms with formal regret guarantees essentially rely on trying all possible behaviors.<n>We show that this combination enables the agent to learn both safely and effectively.<n>Our result may be the first formal proof that it is possible for an agent to obtain high reward while becoming self-sufficient.
• Score: 13.369079495587693
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Most learning algorithms with formal regret guarantees essentially rely on trying all possible behaviors, which is problematic when some errors cannot be recovered from. Instead, we allow the learning agent to ask for help from a mentor and to transfer knowledge between similar states. We show that this combination enables the agent to learn both safely and effectively. Under standard online learning assumptions, we provide an algorithm whose regret and number of mentor queries are both sublinear in the time horizon for any Markov Decision Process (MDP), including MDPs with irreversible dynamics. Our proof involves a sequence of three reductions which may be of independent interest. Conceptually, our result may be the first formal proof that it is possible for an agent to obtain high reward while becoming self-sufficient in an unknown, unbounded, and high-stakes environment without resets.

Related papers

• Agentic Uncertainty Quantification [76.94013626702183]
  We propose a unified Dual-Process Agentic UQ (AUQ) framework that transforms verbalized uncertainty into active, bi-directional control signals.<n>Our architecture comprises two complementary mechanisms: System 1 (Uncertainty-Aware Memory, UAM), which implicitly propagates verbalized confidence and semantic explanations to prevent blind decision-making; and System 2 (Uncertainty-Aware Reflection, UAR), which utilizes these explanations as rational cues to trigger targeted inference-time resolution only when necessary.
  arXiv  Detail & Related papers  (2026-01-22T07:16:26Z)
• Core Safety Values for Provably Corrigible Agents [2.6451153531057985]
  We introduce the first implementable framework for corrigibility, with provable guarantees in multi-step, partially observed environments.<n>Our framework replaces a single reward with five *structurally separate* utility heads.<n>For open-ended settings where adversaries can modify the agent, we prove that deciding whether an arbitrary post-hack agent will ever violate corrigibility is undecidable.
  arXiv  Detail & Related papers  (2025-07-28T16:19:25Z)
• Online inductive learning from answer sets for efficient reinforcement learning exploration [52.03682298194168]
  We exploit inductive learning of answer set programs to learn a set of logical rules representing an explainable approximation of the agent policy.<n>We then perform answer set reasoning on the learned rules to guide the exploration of the learning agent at the next batch.<n>Our methodology produces a significant boost in the discounted return achieved by the agent, even in the first batches of training.
  arXiv  Detail & Related papers  (2025-01-13T16:13:22Z)
• Recursive Introspection: Teaching Language Model Agents How to Self-Improve [30.086494067593268]
  We develop RISE: Recursive IntroSpEction, an approach for fine-tuning large language models. Our experiments show that RISE enables Llama2, Llama3, and Mistral models to improve themselves with more turns on math reasoning tasks.
  arXiv  Detail & Related papers  (2024-07-25T17:35:59Z)
• Jailbreaking as a Reward Misspecification Problem [80.52431374743998]
  We propose a novel perspective that attributes this vulnerability to reward misspecification during the alignment process. We introduce a metric ReGap to quantify the extent of reward misspecification and demonstrate its effectiveness. We present ReMiss, a system for automated red teaming that generates adversarial prompts in a reward-misspecified space.
  arXiv  Detail & Related papers  (2024-06-20T15:12:27Z)
• Multi-Agent Imitation Learning: Value is Easy, Regret is Hard [52.31989962031179]
  We study a multi-agent imitation learning (MAIL) problem where we take the perspective of a learner attempting to coordinate a group of agents. Most prior work in MAIL essentially reduces the problem to matching the behavior of the expert within the support of the demonstrations. While doing so is sufficient to drive the value gap between the learner and the expert to zero under the assumption that agents are non-strategic, it does not guarantee to deviations by strategic agents.
  arXiv  Detail & Related papers  (2024-06-06T16:18:20Z)
• No-Regret Algorithms for Safe Bayesian Optimization with Monotonicity Constraints [41.04951588017592]
  We consider the problem of sequentially maximizing an unknown function $f$ over a set of actions of the form $(s,mathbfx)$. We show that a modified version of our algorithm is able to attain sublinear regret for the task of finding a near-optimal $s$ corresponding to every $mathbfx$.
  arXiv  Detail & Related papers  (2024-06-05T13:41:26Z)
• Learning Constrained Markov Decision Processes With Non-stationary Rewards and Constraints [34.7178680288326]
  In constrained Markov decision processes (CMDPs) with adversarial rewards and constraints, a well-known impossibility result prevents any algorithm from attaining sublinear regret and sublinear constraint violation. We show that this negative result can be eased in CMDPs with non-stationary rewards and constraints, by providing algorithms whose performances smoothly degrade as non-stationarity increases.
  arXiv  Detail & Related papers  (2024-05-23T09:48:48Z)
• Truly No-Regret Learning in Constrained MDPs [61.78619476991494]
  We propose a model-based primal-dual algorithm to learn in an unknown CMDP. We prove that our algorithm achieves sublinear regret without error cancellations.
  arXiv  Detail & Related papers  (2024-02-24T09:47:46Z)
• Information-Theoretic Safe Bayesian Optimization [59.758009422067005]
  We consider a sequential decision making task, where the goal is to optimize an unknown function without evaluating parameters that violate an unknown (safety) constraint. Most current methods rely on a discretization of the domain and cannot be directly extended to the continuous case. We propose an information-theoretic safe exploration criterion that directly exploits the GP posterior to identify the most informative safe parameters to evaluate.
  arXiv  Detail & Related papers  (2024-02-23T14:31:10Z)
• Avoiding Catastrophe in Online Learning by Asking for Help [7.881265948305421]
  We propose an online learning problem where the goal is to minimize the chance of catastrophe.<n>We first show that in general, any algorithm either constantly queries the mentor or is nearly guaranteed to cause catastrophe.<n>We provide an algorithm whose regret and rate of querying the mentor both approach 0 as the time horizon grows.
  arXiv  Detail & Related papers  (2024-02-12T21:12:11Z)
• ReST meets ReAct: Self-Improvement for Multi-Step Reasoning LLM Agent [50.508669199496474]
  We develop a ReAct-style LLM agent with the ability to reason and act upon external knowledge. We refine the agent through a ReST-like method that iteratively trains on previous trajectories. Starting from a prompted large model and after just two iterations of the algorithm, we can produce a fine-tuned small model.
  arXiv  Detail & Related papers  (2023-12-15T18:20:15Z)
• Online Decision Mediation [72.80902932543474]
  Consider learning a decision support assistant to serve as an intermediary between (oracle) expert behavior and (imperfect) human behavior. In clinical diagnosis, fully-autonomous machine behavior is often beyond ethical affordances.
  arXiv  Detail & Related papers  (2023-10-28T05:59:43Z)
• MURAL: Meta-Learning Uncertainty-Aware Rewards for Outcome-Driven Reinforcement Learning [65.52675802289775]
  We show that an uncertainty aware classifier can solve challenging reinforcement learning problems. We propose a novel method for computing the normalized maximum likelihood (NML) distribution. We show that the resulting algorithm has a number of intriguing connections to both count-based exploration methods and prior algorithms for learning reward functions.
  arXiv  Detail & Related papers  (2021-07-15T08:19:57Z)
• Learning to Act Safely with Limited Exposure and Almost Sure Certainty [1.0323063834827415]
  This paper aims to put forward the concept that learning to take safe actions in unknown environments, even with probability one guarantees, can be achieved without the need for exploratory trials. We first focus on the canonical multi-armed bandit problem and seek to study the intrinsic trade-offs of learning safety in the presence of uncertainty.
  arXiv  Detail & Related papers  (2021-05-18T18:05:12Z)
• Robust Asymmetric Learning in POMDPs [24.45409442047289]
  Existing approaches for imitation learning have a serious flaw: the expert does not know what the trainee cannot see. We derive an objective to train the expert to maximize the expected reward of the imitating agent policy, and use it to construct an efficient algorithm, adaptive asymmetric DAgger (A2D) We show that A2D produces an expert policy that the agent can safely imitate, in turn outperforming policies learned by imitating a fixed expert.
  arXiv  Detail & Related papers  (2020-12-31T11:46:51Z)
• Towards Safe Policy Improvement for Non-Stationary MDPs [48.9966576179679]
  Many real-world problems of interest exhibit non-stationarity, and when stakes are high, the cost associated with a false stationarity assumption may be unacceptable. We take the first steps towards ensuring safety, with high confidence, for smoothly-varying non-stationary decision problems. Our proposed method extends a type of safe algorithm, called a Seldonian algorithm, through a synthesis of model-free reinforcement learning with time-series analysis.
  arXiv  Detail & Related papers  (2020-10-23T20:13:51Z)
• Excursion Search for Constrained Bayesian Optimization under a Limited Budget of Failures [62.41541049302712]
  We propose a novel decision maker grounded in control theory that controls the amount of risk we allow in the search as a function of a given budget of failures. Our algorithm uses the failures budget more efficiently in a variety of optimization experiments, and generally achieves lower regret, than state-of-the-art methods.
  arXiv  Detail & Related papers  (2020-05-15T09:54:09Z)
• Corruption-robust exploration in episodic reinforcement learning [76.19192549843727]
  We study multi-stage episodic reinforcement learning under adversarial corruptions in both the rewards and the transition probabilities of the underlying system. Our framework yields efficient algorithms which attain near-optimal regret in the absence of corruptions. Notably, our work provides the first sublinear regret guarantee which any deviation from purely i.i.d. transitions in the bandit-feedback model for episodic reinforcement learning.
  arXiv  Detail & Related papers  (2019-11-20T03:49:13Z)

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.