Tackles bridging machine learning and control theory for safety-critical systems.
problem Ensuring reliability and safety in machine learning applications for safety-critical systems.
method Review of recent advances in learning and control theory, historical context.
result Importance of control theorists joining the conversation on learning-related problems.
Safe RL for safety-critical tasks using CBFs and model-free RL.
problem Limited safety guarantees in RL for real-world applications.
method Combines model-free RL with CBFs and GP dynamics modeling.
result Demonstrates greater sample efficiency and safety in learning.
Predicts failure of autonomous vehicle steering control models.
problem Evaluating and predicting failure of machine learning models in safety-critical applications.
method Trains a student model to predict the main model's error based on saliency maps.
result Preliminary results show the failure predictor model works on autonomous vehicle steering control systems.
UAIL uses uncertainty estimation to improve control systems in safety-critical tasks.
problem Improving control systems in safety-critical domains like autonomous driving.
method UAIL applies Monte Carlo Dropout to estimate uncertainty in control output and selectively acquire new training data.
result UAIL can reliably predict infractions and outperforms existing algorithms.
Paper reviews robustness in machine learning models and discusses training and certification methods.
problem Ensuring reliability of machine learning models in safety-critical systems.
method Reviews formalisms and discusses training and certification techniques.
result Identifies future research directions in robust machine learning.
Meta-active learning optimizes control of safety-critical systems by efficiently learning dynamics and configurations.
problem Efficiently learning system dynamics and optimal configurations for safety-critical systems like deep brain stimulation.
method Meta-learning an acquisition function using LSTM, cast as meta-learning, with a mixed-integer linear program policy.
result Achieved a 46% increase in information gain and a 20% speedup in computation time over baselines.
New method improves reliability of object detection models.
problem Lack of reliability in neural networks for safety-critical applications.
method Sequential Conformal Risk Control (SeqCRC) for object detection.
result Valid statistical guarantees for any dataset size without prior knowledge.
Neural Networks (NN) have been proposed in the past as an effective means for both modeling and control of systems with very complex dynamics. However, despite the extensive research, NN-based controllers have not been adopted by the industry for safety critical systems. The primary reason is that systems with learning…
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.
Paper derives uniform error bounds for Gaussian process regression for safer control applications.
problem Quantifying model error in Gaussian process regression for safety-critical applications.
method Employing Gaussian process distribution and continuity arguments, derive uniform error bounds under weaker assumptions.
result Derives novel uniform error bounds for Gaussian process regression under weaker assumptions.
Reinforcement Learning (RL) is a widely employed machine learning architecture that has been applied to a variety of control problems. However, applications in safety-critical domains require a systematic and formal approach to specifying requirements as tasks or goals. We propose a model-free RL algorithm that enables…
Safe learning of stochastic dynamics with safety constraints.
problem Learning controlled stochastic dynamics with safety constraints.
method Iterative expansion of a safe control set using kernel-based confidence bounds.
result The method ensures safe exploration and efficient estimation of system dynamics.
Paper proposes method for optimal control of unknown systems with latent states.
problem Jointly estimating dynamics and latent states in systems with unmeasurable states.
method Combination of particle Markov chain Monte Carlo methods and scenario theory.
result Probabilistic performance guarantees for optimal input trajectories.
Paper improves neural network robustness analysis for safety-critical systems.
problem Uncertainty in neural network outputs for safety-critical systems.
method Unified propagation and partition approaches to provide tighter bounds.
result Proposed algorithms give tighter bounds than existing methods for the same computation time.
Formal constraints improve RL safety in complex environments.
problem Safety constraints in reinforcement learning for complex environments.
method Specify constraints in formal languages, instantiate as finite automata, augment MDP states, learn dense cost function.
result Improved safety in training RL algorithms over various constraints.
Paper optimizes traffic signal control for better traffic flow.
problem Optimizing traffic signal control to reduce congestion and improve safety.
method Value-based reinforcement learning with interpretable policy functions (polynomial functions).
result Deep Regulatable Hardmax Q-learning variant reduces vehicle delay by up to 19.4%.
Abstract Neural Networks (ANNs) improve DNN verification efficiency.
problem Efficiently verify safety-critical DNNs without slowing exponentially.
method Introduces ANNs that use abstract domains and activation functions to overapproximate DNNs.
result ANNs can soundly overapproximate DNNs with fewer nodes, improving verification efficiency.
Generates multimodal safety-critical scenarios for robustness evaluation of decision-making algorithms.
problem Lack of comprehensive evaluation of neural network robustness under real-world scenarios.
method Proposes a flow-based multimodal scenario generator using weighted likelihood maximization and gradient-based sampling.
result Demonstrates improved testing efficiency and multimodal modeling capability compared to traditional methods.
New bounds for kernel regression under non-Gaussian noise.
problem Uncertainty quantification for function estimates from noisy observations.
method Novel non-asymptotic probabilistic uniform error bounds for kernel-based regression.
result Proposed bounds apply to a broad class of non-Gaussian noise distributions.
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…
Automated testing framework finds weaknesses in deep control policies.
problem Safety of deep neural network control policies is difficult to validate.
method Adversarial reinforcement learning to test and find weaknesses.
result Framework finds weaknesses not evident during manual testing.
Decomposes epistemic uncertainty into per-class contributions for safer classification.
problem Asymmetric costs in safety-critical classification.
method Decomposes mutual information into per-class vector Ck using second-order Taylor expansion. result Decomposition improves selective risk by 34.7% and 56.2% over existing metrics.
ARTEO algorithm optimizes safety-critical systems with uncertainty.
problem Decision-making under uncertainty with safety constraints in real-time optimization.
method ARTEO algorithm uses multi-armed bandits as a mathematical programming problem subject to safety constraints, learning unknown characteristics through exploration and incorporating uncertainty quantification.
result ARTEO achieves less cumulative regret with accurate and safe decisions.
Deep-PrAE improves rare-event simulation for black-box systems.
problem Evaluating rare safety-critical events in learning-based systems.
method Combines deep neural networks with IS to create statistically guaranteed estimations.
result Deep-PrAE provides accurate bounds on safety-critical event probabilities.
Modern vision-based reinforcement learning techniques often use convolutional neural networks (CNN) as universal function approximators to choose which action to take for a given visual input. Until recently, CNNs have been treated like black-box functions, but this mindset is especially dangerous when used for control…
The paper studies how neural policies can be interpreted using decision trees.
problem Understanding how machine learning controllers make decisions in complex environments.
method The approach involves disentangled representation using decision trees to interpret neural policies.
result The paper shows that disentanglement of learned neural dynamics improves interpretability.
DriveFI uses ML to find critical faults in AVs, saving time and resources.
problem Lack of end-to-end fault assessment in AVs under realistic scenarios.
method Machine learning-based fault injection engine (DriveFI) that identifies safety-critical faults.
result Found 561 safety-critical faults in less than 4 hours, compared to weeks of random injection.
Deep reinforcement learning system improves air traffic control efficiency and safety.
problem High-density, dynamic, and stochastic air traffic control challenges.
method Deep multi-agent reinforcement learning framework using actor-critic model with PPO loss function.
result Framework resolves 99.97% of all conflicts in extreme high-density scenarios.
HardNet adds hard constraints to neural networks without sacrificing performance.
problem Ensuring adherence to input-dependent constraints in neural networks.
method Appends a differentiable enforcement layer to neural networks for end-to-end training with hard constraint guarantees.
result HardNet retains neural networks' universal approximation capabilities and enables efficient optimization.
A framework identifies worst-case decision points in safety-critical scenarios, improving risk assessment by 10 hours.
problem Identifying worst-case outcomes in safety-critical decision-making under uncertainty.
method Explicitly estimating distributions of expected return to identify dead-ends, tuning based on risk tolerance.
result Significantly improves risk assessment, providing indications 10 hours earlier and increasing detection by 20%.
SGPA calibrates transformer uncertainty for safety-critical tasks.
problem Uncertainty estimation in transformer models for safety-critical domains.
method Bayesian inference in transformer's output space using sparse Gaussian processes.
result SGPA-based Transformers improve in-distribution calibration and out-of-distribution robustness.
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…
Develops a method to simulate rare dangerous events in autonomous systems.
problem Rare dangerous events in safety-critical systems are hard to test in real-world settings.
method Combines exploration, exploitation, and optimization techniques for rare-event simulation.
result Provides rigorous guarantees for the performance of the method.
The paper tackles MAP inference over non-convex constraints in safety-critical settings.
problem Efficiently computing MAP predictions subject to non-convex constraints is challenging.
method The paper investigates conditions for exact and efficient MAP inference over continuous variables and devises scalable algorithms for both tractable and general cases.
result The proposed methods outperform constraint-agnostic baselines and scale to complex densities.
Semantic embeddings improve safety-critical classifier performance.
problem Improving interpretability and error detection in safety-critical neural networks.
method Created embeddings from symbolic domain knowledge, used for misprediction interpretation and error detection, introduced semantic distance for confidence measurement.
result Semantic distance achieves near state-of-the-art performance in a traffic sign classifier, faster than other methods.
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.
Recent successes in reinforcement learning have lead to the development of complex controllers for real-world robots. As these robots are deployed in safety-critical applications and interact with humans, it becomes critical to ensure safety in order to avoid causing harm. A first step in this direction is to test the …
Fatal accidents are a major issue hindering the wide acceptance of safety-critical systems using machine-learning and deep-learning models, such as automated-driving vehicles. Quality assurance frameworks are required for such machine learning systems, but there are no widely accepted and established quality-assurance …
ViTaX provides formal guarantees for targeted explanations in safety-critical systems.
problem Need trustworthy explanations for safety-critical deep neural networks.
method Formal reachability analysis for targeted, semifactual explanations.
result First method to provide formally guaranteed explanations of model resilience.
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.
New RL framework improves real-time control performance.
problem Real-time RL systems assume static states, leading to suboptimal outcomes.
method Introduces a new real-time RL framework where states and actions evolve simultaneously.
result RTAC algorithm outperforms existing state-of-the-art algorithms in real-time and non-real-time settings.
DL models for MTS regression are vulnerable to adversarial attacks, posing risks in safety-critical applications.
problem Vulnerability of DL models to adversarial examples in MTS regression.
method Adversarial attack generation techniques from image classification were adapted for MTS.
result All state-of-the-art DL regression models (CNN, LSTM, GRU) are vulnerable to adversarial attacks.
The paper investigates how to ensure safety-critical reinforcement learning agents perform well in novel situations.
problem Ensuring safety-critical reinforcement learning agents perform well in unseen test environments.
method Investigates safety and generalization from a limited number of training environments, using gridworld and CoinRun settings.
result Simple modifications like ensemble model averaging and blocking classifier can reduce catastrophes in gridworld, but not in CoinRun.
SkyGP improves Gaussian process scalability for real-time learning.
problem Scalability issues with exact Gaussian processes for streaming data.
method Streaming kernel-induced progressively generated Gaussian process experts (SkyGP).
result SkyGP maintains performance guarantees while improving scalability.
This paper develops a framework for training and evaluating neural networks for MPC.
problem Lack of a general framework for characterizing learning approaches in MPC.
method Developed a framework using PyTorch and CVXPY, incorporating hit-and-run sampling for efficient training data generation.
result Proposed metrics for validating neural network-based MPC approaches.
Adversarial attacks pose a threat to deep neural networks, especially in safety-critical applications.
problem Adversarial attacks can misclassify deep neural networks, leading to safety issues.
method Adversarial attacks are categorized into white-box and black-box attacks based on the attacker's knowledge. They can be targeted or non-targeted.
result Adversarial attacks are effective and can transfer between different models and real-world scenarios.
The paper tackles safe exploration in RL by a conservative safety critic.
problem Safe exploration in reinforcement learning (RL) when partially trained policies are deployed.
method Learning a conservative safety estimate through a critic, provably bounding catastrophic failures.
result The approach provably converges to competitive task performance with significantly lower catastrophic failure rates.
Quantum model improves safety in machine learning.
problem Improving safety and robustness in machine learning models.
method Variational quantum classifier with amplitude encoding and SAFE-AI metrics.
result Quantum model provides competitive performance and improved robustness.