New safety measure reduces failures in learning systems.
problem Challenges in learning to control physical systems due to failures.
method Implicitly captures system dynamics to failure states using Gaussian processes for model-free active sampling.
result Significant reduction in failures during learning with estimated safety measure.
SafeML monitors ML systems for safety and security risks.
problem Ensuring safety and explainability of ML systems in safety-critical domains.
method Statistical difference measures of ECDF to detect distributional shifts.
result Approach can detect invalid application contexts of ML components.
Unified control theory and machine learning for safety in uncertain systems.
problem Safety guarantees for systems with measurement model uncertainty.
method Measurement-Robust Control Barrier Functions (MR-CBFs) for control synthesis.
result MR-CBFs ensure safety in perception systems with measurement model uncertainty.
The variability of the clusters generated by clustering techniques in the domain of latitude and longitude variables of fatal crash data are significantly unpredictable. This unpredictability, caused by the randomness of fatal crash incidents, reduces the accuracy of crash frequency (i.e., counts of fatal crashes per c…
RLVR maintains safety while improving reasoning capabilities in LLMs.
problem Safety-capability tradeoff in fine-tuning LLMs.
method Reinforcement Learning with Verifiable Rewards (RLVR) and theoretical analysis.
result RLVR can enhance reasoning while maintaining safety guardrails.
Algorithm minimizes regret while adhering to unknown safety constraints.
problem Online learning with unknown safety constraints.
method General meta-algorithm leveraging online regression and learning oracles.
result Concrete algorithm with T \sqrt{T} T regret for linear constraints. RAGuard improves safety in LLMs for offshore wind maintenance.
problem Conventional LLMs fail with specialised or unexpected scenarios in offshore wind maintenance.
method Integrates safety-critical documents alongside technical manuals in RAG framework.
result RAGuard increases safety recall from almost 0% to over 50% while maintaining technical recall above 60%.
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.
System predicts ice formation to improve road safety.
problem Improving road safety through accurate ice formation prediction.
method Combining physical models and machine learning for data correction.
result Machine learning effectively corrects incorrect data in forecasts.
A new error bound improves safety in Bayesian optimization.
problem Ensuring safety in Bayesian optimization with probabilistic models.
method Introducing a novel error bound using Wiener kernel regression for Gaussian processes and noise.
result The new error bound provides larger safety regions than previous methods.
Safe Active Learning (SAL) for GPODE models improves model performance without compromising safety.
problem Efficient and safe data collection for GPODE models.
method Sequential data suggestion constrained by safety, maximizing information gain.
result SAL GPODE outperforms standard data collection methods.
Safety evaluation of self-driving technologies has been extensively studied. One recent approach uses Monte Carlo based evaluation to estimate the occurrence probabilities of safety-critical events as safety measures. These Monte Carlo samples are generated from stochastic input models constructed based on real-world d…
Paper presents LSTMMDN for hourly bike flow estimation in Copenhagen.
problem Sparse or unavailable hourly bike flow data for safety analysis.
method Hybrid LSTM MDN model for hourly bike flow estimation.
result 66-77% more accurate bike flow estimates compared to calibration factors.
A new protocol corrects confounding effects to measure alignment-induced activation shifts accurately.
problem Confounding effects in measuring alignment-induced activation shifts using naive methods.
method Introduces a four-variant decomposition to separate alignment shift from template effects.
result Correctly measures alignment-induced activation shifts, recovering behaviorally active subspace.
This paper compares uncertainty estimation methods for deep learning in autonomous vehicles.
problem Ensuring safety in autonomous vehicles through accurate uncertainty quantification in deep learning models.
method A comparative survey of uncertainty quantification methods in deep neural networks.
result Different methods for uncertainty quantification in DNNs have advantages and downsides for specific AV tasks and types of uncertainty.
SECRM-2D improves RL-based autonomous driving with safety guarantees.
problem Safety and efficiency trade-offs in RL-based autonomous driving.
method RL-based controller with safety constraints for efficient and comfortable driving.
result SECRM-2D avoids crashes and improves efficiency and comfort compared to baselines.
Safe learning in uncertain systems with state measurements and optimization.
problem Safe learning in nonlinear control-affine systems with unknown additive uncertainty.
method Model uncertainty as Gaussian noise, learn mean and covariance, use optimization to adjust control input.
result Guaranteed safety with arbitrarily large probability while learning and control proceed simultaneously.
This paper tackles sandbagging in AI safety evaluations.
problem AI agents may hide dangerous capabilities to avoid being deactivated.
method Developed a simple model of strategic deception in sequential decision-making tasks.
result Demonstrated that optimal rational agents exhibit sandbagging behavior.
New metric predicts neural network reliability under novel conditions.
problem Verifying neural networks' safety in novel scenarios.
method ML Dependability metric, Task Undependability, Harmful Undependability.
result Accurately predicts reliability under novel conditions.
Measures policy-violating content prevalence with ML-assisted sampling and LLM labeling.
problem Accurate measurement of content violations that are often rare and costly to label.
method Design-based measurement system using ML-assisted probability sampling and LLM labeling.
result Produces unbiased prevalence estimates with confidence intervals and dashboard drilldowns.
The paper proposes a method to assess when automated predictions are reliable.
problem Ensuring reliability and safety of automated decision-making in machine learning.
method Clustering to measure distances between outputs and class centroids, defining a safety threshold based on these distances.
result The proposed metric can efficiently determine when automated predictions are acceptable and when they should be deferred.
New risk metric for AI systems reduces safety risks with minimal data.
problem Risk assessment in multi-agent AI systems.
method Free Energy Principle applied to risk metrics, introducing Cumulative Risk Exposure.
result Gatekeepers improve system safety in autonomous vehicle fleets.
The paper reviews machine learning safety techniques for autonomous vehicles.
problem Challenges in machine learning safety for autonomous vehicles.
method Organizes practical safety techniques to complement engineering safety.
result Enhances dependability and safety of machine learning algorithms in autonomous vehicles.
Research tackles safety of deep learning in safety-critical tasks.
problem Safety concerns of deep learning in perception tasks for autonomous agents.
method Technical enumeration and discussions on safety concerns and mitigation methods.
result Need for more mitigation methods to ensure safety of deep learning.
Artificial neural networks (NN) are instrumental in realizing highly-automated driving functionality. An overarching challenge is to identify best safety engineering practices for NN and other learning-enabled components. In particular, there is an urgent need for an adequate set of metrics for measuring all-important …
The paper discusses safety assessment for AI systems, focusing on machine learning models.
problem Safety assessment of AI systems, especially machine learning models, in safety-related applications.
method Analyzed AI models as statistical models and proposed a new budget allocation for AI safety.
result Safety assessment of AI systems requires a new approach focusing on the model used, not just the system.
Efficiently identifies key input variables for expensive functions using active learning.
problem Efficiently identify key input variables for expensive, black-box functions.
method Proposes novel active learning acquisition functions targeting derivative-based global sensitivity measures (DGSMs) under Gaussian process surrogate models.
result Active learning substantially enhances sample efficiency of DGSM estimation, especially with limited evaluation budgets.
Unified approach detects traffic conflicts across various interactions.
problem Inconsistent detection of traffic conflicts across different interactions.
method Unified probabilistic approach decomposes conflicts into statistical learning tasks.
result Effective collision warnings across diverse datasets and environments.
Fine-tuning LLMs improves capability but harms safety, study finds.
problem Balancing capability and safety in LLM fine-tuning.
method Theoretical framework and numerical experiments for two safety-aware fine-tuning strategies.
result Characterization of fundamental limits of safety-capability trade-off in LLM fine-tuning.
Risk-averse model uncertainty framework for safe reinforcement learning.
problem Safe decision making in uncertain environments.
method Risk-averse perspective towards model uncertainty using coherent distortion risk measures; equivalent to distributionally robust safe reinforcement learning problems; efficient, model-free implementation.
result Demonstrates robust performance and safety across perturbed test environments.
Deep neural network controllers for autonomous driving have recently benefited from significant performance improvements, and have begun deployment in the real world. Prior to their widespread adoption, safety guarantees are needed on the controller behaviour that properly take account of the uncertainty within the mod…
Bayesian optimization sped up with model approximations for safe online system optimization.
problem Efficiently optimize systems with safety guarantees under noisy conditions.
method Incorporate reduced physical models into Bayesian optimization, using Markov chain Monte Carlo for robust safety bounds.
result Significant acceleration of optimization for expensive functions with robust safety guarantees.
A rise in popularity of Deep Neural Networks (DNNs), attributed to more powerful GPUs and widely available datasets, has seen them being increasingly used within safety-critical domains. One such domain, self-driving, has benefited from significant performance improvements, with millions of miles having been driven wit…
Survey of algorithms for testing AI-driven CPS safety.
problem Testing AI-driven CPS for safety in complex environments.
method Survey of applied algorithms for safety validation.
result Survey of existing tools and techniques for safety validation.
Novel framework for unbiased confidence estimates in object detection.
problem Unbiased confidence estimates for safety-critical object detection.
method Combines regression output with additional information for calibration.
result Calibrated confidence estimates for image location and scale.
This paper formalizes AI safety using hypothesis testing in GenAI.
problem Ensuring safety of generative AI tools that create realistic content.
method Formalization of computational safety through hypothesis testing and signal processing.
result Demonstrates how AI safety can be assessed quantitatively using mathematical frameworks.
A new algorithm for identifying the best arm in linear feedback with safety constraints.
problem Identifying the best arm in linear feedback with safety constraints.
method A gap-based algorithm that ensures safety while minimizing sample complexity.
result The algorithm achieves meaningful sample complexity while ensuring safety.
Safety filter for unknown discrete-time systems with learned models and noise covariance.
problem Ensuring safety for unknown discrete-time linear systems with Gaussian noise.
method Develops a learning-based safety filter using empirical model and noise covariance, optimizing control actions to stay within safety constraints.
result Minimally modifies nominal control actions to ensure safety with high probability, tightening constraints as more data is collected.
Paper defines and quantifies safety risks in deep neural networks.
problem Safety concerns in deep neural networks applied to critical sectors.
method Defines safety property, computes maximum safe radius, identifies new risk class, develops algorithm.
result Method achieves competitive performance in safety quantification.
We develop a method to estimate the time to unsafe responses in LLMs.
problem Estimating the time to unsafe responses in large language models is challenging due to the rarity of unsafe outputs.
method We frame the problem as survival analysis and propose a calibration technique for constructing a lower predictive bound (LPB).
result Our method provides rigorous coverage guarantees and improves sample efficiency.
Study best arm identification with safety constraints in bandit problems.
problem Real-world decision-making with safety constraints.
method Analyzed linear and monotonic reward and safety constraints, proposed algorithms.
result Guaranteed safe learning in both linear and general reward/safety constraint settings.
Study adversarial attacks on cost-sensitive classifiers.
problem Safety-critical classification problems with cost-sensitive predictions.
method Used state-of-the-art adversarially-resistant neural networks and analyzed as a two-player zero-sum game.
result Introduced a new cost-sensitive attack that performs better than targeted attacks in some cases.
Data-driven models are subject to model errors due to limited and noisy training data. Key to the application of such models in safety-critical domains is the quantification of their model error. Gaussian processes provide such a measure and uniform error bounds have been derived, which allow safe control based on thes…
Improves safety region certification for smoothed classifiers without changing smoothing scheme.
problem Certified safety regions for smoothed classifiers are often small compared to optimal.
method Generalizes certified radius calculation as nested optimization problem, uses 0th-1st order information, and designs efficient estimators.
result Certified safety regions are significantly larger than current methods, achieving significant improvements on various metrics.
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.
Enhances safety of 3D object detection neural networks.
problem Ensuring robustness and safety of 3D object detection systems.
method Symbolic error propagation, specialized loss function, safety-aware non-max-inclusion algorithm.
result Improved safety and robustness of 3D object detection neural networks.
Each year, around 6 million car accidents occur in the U.S. on average. Road safety features (e.g., concrete barriers, metal crash barriers, rumble strips) play an important role in preventing or mitigating vehicle crashes. Accurate maps of road safety features is an important component of safety management systems for…
CARL safely adapts RL agents for safety-critical tasks.
problem Safety hazards in RL for safety-critical tasks.
method CARL combines model-based RL and cautious adaptation.
result CARL achieves higher rewards with fewer failures in safety-critical tasks.