Algorithm reduces regret in safe Bayesian optimization with monotonicity constraints.
problem Sequentially maximize unknown function with safety constraints.
method Sequential algorithms using Gaussian processes with safety constraints modeled as monotonicity.
result Sublinear regret achieved for expanding safe region and finding optimal s. Bandit algorithms have various application in safety-critical systems, where it is important to respect the system constraints that rely on the bandit's unknown parameters at every round. In this paper, we formulate a linear stochastic multi-armed bandit problem with safety constraints that depend (linearly) on an unkn…
Safe imitation learning with a safety layer for flexible training.
problem Flexible yet safe imitation learning for complex tasks.
method Theory and modular method with a safety layer for continuous policy, adversarial training, and worst-case safety guarantees.
result Robustness advantage of safety layer during training compared to test time.
Safe actions learned in finite trials, without infinite exploration.
problem Learning safe actions in unknown environments efficiently.
method Defining a handicap metric and using sequential probability ratio test for discarding unsafe actions.
result Achieves constant handicap, discarding unsafe machines with probability one in finite rounds.
Proposes a method to accelerate safe sequential learning using offline data.
problem Limited exploration due to disconnected safe regions and slow task learning.
method Safe transfer sequential learning using Gaussian processes and offline data.
result Enhances global exploration across multiple disjoint safe regions with lower data consumption.
In the real world, agents often have to operate in situations with incomplete information, limited sensing capabilities, and inherently stochastic environments, making individual observations incomplete and unreliable. Moreover, in many situations it is preferable to delay a decision rather than run the risk of making …
Safe linear bandits over unknown polytopes avoid safety violations and suboptimal actions.
problem Online approach to linear programming with unknown constraints and risks.
method Doubly-optimistic strategy (DOSS) for safe linear bandits over polytopes.
result DOSS achieves tight bounds on efficacy regret and safety violations.
In classical reinforcement learning, when exploring an environment, agents accept arbitrary short term loss for long term gain. This is infeasible for safety critical applications, such as robotics, where even a single unsafe action may cause system failure. In this paper, we address the problem of safely exploring fin…
This paper proposes a method to safely adjust exploration in RL to satisfy constraints.
problem Unsafe exploration in reinforcement learning violates constraints on controlled object states.
method Automatic adjustment of exploration inputs and variance-covariance matrix for safety.
result The method guarantees satisfaction of joint chance constraints with specified probability.
Safe RL with binary feedback using SABRE algorithm.
problem Safe reinforcement learning with binary safety feedback.
method SABRE algorithm, combining active learning and reinforcement learning.
result Provable safe policy with high probability, no unsafe actions during training.
In reinforcement learning, agents learn by performing actions and observing their outcomes. Sometimes, it is desirable for a human operator to \textit{interrupt} an agent in order to prevent dangerous situations from happening. Yet, as part of their learning process, agents may link these interruptions, that impact the…
We introduce the safe linear stochastic bandit framework---a generalization of linear stochastic bandits---where, in each stage, the learner is required to select an arm with an expected reward that is no less than a predetermined (safe) threshold with high probability. We assume that the learner initially has knowledg…
Safe Gaussian Process Bandit Optimization with sub-linear regret bounds.
problem Sequential decision-making under uncertainty and safety constraints.
method Developed SGP-UCB, a safe variant of GP-UCB with modifications to respect safety constraints.
result First sub-linear regret bounds for safe Gaussian Process Bandit Optimization.
Safe-DRFS selects features robust to covariate shifts for reliable performance.
problem Feature selection fails in diverse deployment environments.
method Safe-DRFS extends safe screening to distributionally robust settings under covariate shift.
result Safe-DRFS identifies a feature subset encompassing optimal subsets across distribution shifts.
Safe screening improves generalized CGM's feature selection stability.
problem Improving feature selection stability in generalized CGM.
method Coupling safe screening with generalized CGM.
result Safe screening matches solution support at rate O(1/(tδ2)). Safe screening rule improves Group SLOPE efficiency.
problem Efficiently selecting groups of predictors in high-dimensional sparse learning.
method Safe screening rule for Group SLOPE, addressing block non-separable group effects.
result Significant computational efficiency gains without sacrificing accuracy.
Insurance contracts for autonomous AI agents must be actuarially sound and resistant to gaming.
problem Designing insurance contracts for autonomous AI agents that are actuarially sound and resistant to gaming.
method Characterizing a five-attack space and proving the actuarial runtime is gaming-resistant.
result An incentive-compatible layer for actuarial control of autonomous-agent side effects.
We study the problem of safe learning and exploration in sequential control problems. The goal is to safely collect data samples from operating in an environment, in order to learn to achieve a challenging control goal (e.g., an agile maneuver close to a boundary). A central challenge in this setting is how to quantify…
Revel tackles safe exploration in RL with verified symbolic policies.
problem Computational infeasibility of verifying neural networks in RL learning loops.
method Two policy classes: neurosymbolic with approximate gradients and symbolic policies for efficient verification. Mirror descent over policies to safely update and project policies.
result Revel discovers policies that outperform prior approaches to verified exploration.
ASE safely explores unknown MDPs with unknown dynamics, improving sample efficiency.
problem Balancing exploration and safety in unknown MDPs with stochastic dynamics.
method Exploits analogies between state-action pairs to safely learn near-optimal policies.
result Empirically improves sample efficiency compared to existing methods.
Sparse learning techniques have been routinely used for feature selection as the resulting model usually has a small number of non-zero entries. Safe screening, which eliminates the features that are guaranteed to have zero coefficients for a certain value of the regularization parameter, is a technique for improving t…
The operational space of an autonomous vehicle (AV) can be diverse and vary significantly. This may lead to a scenario that was not postulated in the design phase. Due to this, formulating a rule based decision maker for selecting maneuvers may not be ideal. Similarly, it may not be effective to design an a-priori cost…
Safe Pattern Pruning reduces pattern explosion in predictive pattern mining.
problem Exponential growth of patterns in structured data.
method Safe Pattern Pruning (SPP) method.
result Effective model building in practical data analysis.
This work improves motion planning for quadcopters by learning and reasoning about controller performance.
problem Improving motion planning for quadcopters with safety margins and execution reliability.
method Introspective learning and reasoning to correct execution bias and improve collision checking.
result Substantial reduction in safety margins for motion actions, leading to safer execution.
New attack manipulates UCB algorithm, new defense algorithm reduces pseudo-regret.
problem Adversarial attacks on stochastic bandit algorithms.
method Introducing action-manipulation attacks and proposing a robust defense algorithm.
result Proposed defense algorithm reduces pseudo-regret to O(max{log T, A}).
Safe screening rules reduce ℓ0-regression computation by fixing 76% of variables.
problem Efficiently solving ℓ0-regression problems with large datasets. method Convex relaxation and safe screening rules to eliminate variables.
result 76% of variables can be fixed to their optimal values, reducing computational burden.
Safe offline RL for chemical reactors using input convex neural networks.
problem Safe control of exothermic polymerization reactors using historical data.
method Gymnasium-compatible simulation, behaviour cloning, implicit Q-learning, input convex neural networks (PICNNs).
result Offline RL with convex action correction outperforms traditional control approaches.
Proposes Constrained Q-learning for reinforcement learning with constraints.
problem Optimizing multiple objectives while adhering to constraints in reinforcement learning.
method Directly restricts the action space in Q-update to learn optimal Q-function for constrained MDP.
result Improves safety and optimality in high-level decision making for autonomous driving.
Safe Bayesian optimization method using information theory.
problem Optimizing unknown functions while respecting safety constraints.
method Information-theoretic exploration criterion for continuous domains.
result The method learns the value of the safe optimum up to arbitrary precision.
New method provides scalable safety guarantees for RL agents.
problem Safe reinforcement learning in real-life scenarios.
method State-augmentation and shield design for probabilistic avoidance.
result Strict formal safety guarantees for RL agents, scalable and practical.
Batch Reinforcement Learning (Batch RL) consists in training a policy using trajectories collected with another policy, called the behavioural policy. Safe policy improvement (SPI) provides guarantees with high probability that the trained policy performs better than the behavioural policy, also called baseline in this…
We present safe active incremental feature selection~(SAIF) to scale up the computation of LASSO solutions. SAIF does not require a solution from a heavier penalty parameter as in sequential screening or updating the full model for each iteration as in dynamic screening. Different from these existing screening methods,…
The design and performance analysis of bandit algorithms in the presence of stage-wise safety or reliability constraints has recently garnered significant interest. In this work, we consider the linear stochastic bandit problem under additional \textit{linear safety constraints} that need to be satisfied at each round.…
Practical reinforcement learning problems are often formulated as constrained Markov decision process (CMDP) problems, in which the agent has to maximize the expected return while satisfying a set of prescribed safety constraints. In this study, we propose a novel simulator-based method to approximately solve a CMDP pr…
Previous work has shown the unreliability of existing algorithms in the batch Reinforcement Learning setting, and proposed the theoretically-grounded Safe Policy Improvement with Baseline Bootstrapping (SPIBB) fix: reproduce the baseline policy in the uncertain state-action pairs, in order to control the variance on th…
We consider the problem of learning a sparse rule model, a prediction model in the form of a sparse linear combination of rules, where a rule is an indicator function defined over a hyper-rectangle in the input space. Since the number of all possible such rules is extremely large, it has been computationally intractabl…
Framework trains safe agents avoiding deceptive behavior.
problem Training safe agents from unsafe incentives.
method Formal settings, causal influence analysis, maximizing non-mediated effects.
result Agents avoid manipulating delicate state for rewards.
Monotonicity helps in safely optimizing unknown functions.
problem Sequentially maximizing an unknown function with safety constraints.
method Gaussian process with monotonicity assumption, inspired by GP-UCB and SafeOpt.
result The proposed M-SafeUCB algorithm achieves theoretical guarantees and safety.
We study continuous action reinforcement learning problems in which it is crucial that the agent interacts with the environment only through safe policies, i.e.,~policies that do not take the agent to undesirable situations. We formulate these problems as constrained Markov decision processes (CMDPs) and present safe p…
Reinforcement learning is a powerful paradigm for learning optimal policies from experimental data. However, to find optimal policies, most reinforcement learning algorithms explore all possible actions, which may be harmful for real-world systems. As a consequence, learning algorithms are rarely applied on safety-crit…
A new algorithm trains experts to safely guide agents in partially observed environments.
problem Existing imitation learning methods for POMDPs can lead to sub-optimal or unsafe policies.
method Derive an objective to encourage the expert to maximize the agent's reward, then use it to train both expert and agent.
result The algorithm produces an expert policy that the agent can safely imitate, outperforming fixed expert policies.
In this work we seek for an approach to integrate safety in the learning process that relies on a partly known state-space model of the system and regards the unknown dynamics as an additive bounded disturbance. We introduce a framework for safely learning a control strategy for a given system with an additive disturba…
Aims to create safe reinforcement learning policies by considering individual harm.
problem Optimal policies for a population may harm certain individuals.
method Formalizes individual harm, proposes a two-stage procedure, and establishes finite-sample properties.
result Learned policies maximize expected return while minimizing harm.
Safe exploration in RF-RL doesn't increase sample complexity.
problem Achieving optimal policies with safety constraints in reward-free RL.
method Proposed SWEET framework for tabular and low-rank MDP settings, leveraging truncated value functions.
result Sample complexities match or outperform constraint-free counterparts, proving safety constraints have little impact.
A new runtime for AI agents calculates risks in real-time.
problem Managing risks and liabilities in autonomous AI actions.
method A time-consistent counterfactual actuarial layer with explicit underwriting boundaries.
result Establishes a well-defined toll and guarantees executed-action budgets.
Safe reinforcement learning tackles safety constraints with linear approximations.
problem Ensuring safety in reinforcement learning without violating constraints.
method Modeling safety as a linear cost function, developing SLUCB-QVI and RSLUCB-QVI algorithms for MDPs with linear function approximation.
result Achieved a nearly optimal regret bound for safe reinforcement learning, matching state-of-the-art unsafe algorithms.
Paper proposes Vertex Networks for reinforcement learning of control systems with safety guarantees.
problem Challenges in reinforcement learning with hard state and action constraints.
method Vertex Networks incorporate safety constraints into policy network architecture, ensuring safety during exploration.
result Proposed Vertex Networks outperform vanilla reinforcement learning in benchmark control tasks.
The paper tackles safe reinforcement learning with convex regularization.
problem Safe reinforcement learning in complex, high-dimensional settings with safety constraints.
method Doubly-regularized RL framework combining reward and parameter regularization, formulated as a convex regularized objective with parametrized policies on an infinite-dimensional statistical manifold.
result Exponential convergence guarantees under sufficient regularization, robust theoretical insights and guarantees for safe RL.