New regularizers tighten convex relaxation bounds for neural networks.
problem Large gap between certifiable and empirical robustness in neural networks.
method Two regularizers to train neural networks yielding tighter convex relaxation bounds.
result Higher certified accuracy with proposed regularizers.
Efficient local Lipschitz bounds improve neural network robustness.
problem Certifying robustness of neural networks is challenging and often leads to over-regularization.
method Proposes an efficient trainable local Lipschitz upper bound by considering activation functions and weight matrices.
result Consistently outperforms state-of-the-art methods in clean and certified accuracy on various datasets.
Paper proposes faster certified robust training methods with short warmup.
problem Certified robust training methods require long warmup schedules, making training costly.
method Proposes three improvements: new weight initialization, BN, and regularization.
result Achieves 65.03% verified error on CIFAR-10 and 82.36% on TinyImageNet with short warmup.
Paper aims to ensure reliable detection of out-of-distribution data with certifiable worst-case guarantees.
problem Deep neural networks are overconfident with OOD inputs, posing safety risks.
method Enforces low confidence and bounds in an l∞-ball around OOD points using interval bound propagation (IBP). result Certifiable worst-case guarantees for OOD detection are possible without significant loss in accuracy.
New work shows limits of certifying neural network robustness.
problem Certified training improves robustness but decreases accuracy.
method Bayes error analysis to investigate robustness limits.
result Upper bound for certified robust accuracy established.
Proves the existence of accurate, certifiably robust neural networks.
problem Training neural networks to be robust against adversarial attacks.
method Proves the existence of networks that approximate continuous functions and ensure robustness through interval-bound propagation.
result Proves the existence of accurate, interval-certified ReLU networks.
New method certifies neural network function space norms from point evaluations.
problem Certifying neural network function space norms from point evaluations alone.
method Combining interval arithmetic enclosures, adaptive marking/refinement, and quadrature-based aggregation.
result Certified computation of Lp, W1,p, and W2,p norms. This work improves certifiably robust models by distilling knowledge from adversarially robust teachers.
problem Certifiably robust models suffer from poor standard performance.
method Knowledge distillation from adversarially robust teachers to improve standard performance.
result Distillation from adversarially robust teachers consistently improves certified training performance.
Certified algorithms optimize functions with varying costs, providing error bounds.
problem Optimizing functions with varying evaluation costs and error bounds.
method Formalized as a min-max game, proposed certified MFDOO algorithm with cost complexity bound.
result Proposed certified MFDOO algorithm has near-optimal cost complexity for Lipschitz functions.
New attack tricks certifiably robust models into mislabeling images.
problem Defeating certified defenses against adversarial examples.
method Spoofed robustness certificates and large perturbations.
result Certifiably robust models can be fooled by large perturbations.
New method certifies global robustness of neural networks efficiently.
problem Adversarial examples threaten certifiably robust neural networks.
method Formalized global robustness, adapted widely-used architectures with efficient global Lipschitz bounds.
result Certifiable robust models achieve state-of-the-art verifiable accuracy with negligible costs.
This paper assesses Gaussian and Exponential mechanisms for certifying adversarial robustness.
problem Certifying adversarial robustness using randomized smoothing mechanisms.
method Proposes a generic framework to assess the appropriateness of randomized smoothing mechanisms.
result Gaussian mechanism is an appropriate option for certifying both ℓ2-norm and ℓ∞-norm robustness. CROWN certifies robustness of neural networks with general activation functions.
problem Certifying robustness of neural networks with general activation functions.
method Bounding activation functions with linear and quadratic surrogates, adaptively selecting surrogates for each neuron.
result Significantly improves certified lower bounds on ReLU networks compared to Fast-Lin.
RobBoost optimizes a deep model ensemble for certified robustness.
problem Improving the robustness of deep models.
method Optimizes a neural network's robustness certificate through model selection and weighting.
result RobBoost forms a more robust ensemble with better certified robustness and clean accuracy.
New method certifies neural network robustness to random input noise.
problem Certifying neural networks' robustness to random input noise.
method Chance-constrained optimization problem reformulated using input-output samples.
result Certifies a uniform infinity-norm uncertainty region with a radius 50 times larger.
CIVET method provides robustness guarantees for VAEs under adversarial attacks.
problem Certified probabilistic guarantees for VAEs in safety-critical applications.
method Bounding worst-case VAE error by error on support sets at the latent layer.
result CIVET outperforms state-of-the-art methods in robustness and standard performance.
New method for certified unlearning reduces noise injection.
problem Achieving formal unlearning guarantees with adaptive noise calibration.
method Adaptive per-instance noise calibration based on individual data point sensitivities.
result Derivation of high-probability per-instance sensitivity bounds for ridge regression.
Certifiable defense method improves robustness of deep learning interpretation.
problem Vulnerability of gradient-based saliency maps to adversarial attacks.
method Sparsified SmoothGrad method, extending certifiably robust smooth classifier bounds.
result Sparsified SmoothGrad method is certifiably robust against adversarial perturbations.
Certifiably robust VAEs are trained with bounds on input perturbations.
problem Ensuring VAEs are robust to adversarial attacks.
method Derive bounds on minimal perturbation size, control parameters, and train VAEs to meet criteria.
result Certifiably robust VAEs are more robust to attacks than standard VAEs.
New framework solves low-rank optimization problems to certifiable optimality.
problem Low-rank optimization problems with certifiable solutions.
method Mixed-Projection Conic Optimization framework using symmetric projection matrices and outer-approximation algorithms.
result Solves low-rank problems to certifiable optimality, outperforming existing methods.
New approach to certifiably robust neural networks using Boolean function perspective.
problem Lack of principled understanding and certified robustness for ℓ∞ perturbations. method New perspective on Boolean functions, deriving impossibility results, and developing a unified Lipschitz network.
result Unified Lipschitz network that bypasses expressive power limitations and achieves better certified robustness.
PROWL uses robust reward estimates to improve ITR selection.
problem Reward uncertainty in ITR estimation leads to inflated performance.
method PAC-Bayesian framework with reward uncertainty certificates.
result PROWL achieves better robust treatment regime estimation.
We certify federated learning model performance under meta-distribution shifts.
problem Certifying model performance on unseen networks with heterogeneous distributions.
method Derive worst-case uniform guarantees for federated learning model's average loss and risk CDF.
result Asymptotically minimax optimal and privacy-preserving certification.
New method certifies images against transformations like rotations and translations.
problem Certifying robustness of images against transformations like rotations and translations.
method Randomized smoothing with three different kinds of defenses.
result Individual certificates can be obtained via statistical error bounds or efficient online inverse computation.
New method solves matrix completion problems to certifiable optimality.
problem Certifying optimality in low-rank matrix completion.
method Disjunctive branch-and-bound scheme for convex relaxation.
result Decreases optimality gap by two orders of magnitude.
This work certifies non-uniform bounds against adversarial attacks for neural networks.
problem Certifying robust regions around data points against non-uniform adversarial attacks.
method Formulated as an optimization problem with nonlinear constraints, using the augmented Lagrangian method for general feedforward neural networks.
result Non-uniform bounds have larger volumes and better interpretability compared to uniform bounds.
New framework tightens certified robustness gaps in machine learning models.
problem Persistent gap between theoretical certified robustness and empirical accuracy.
method Leverages Lipschitz continuity and novel confidence intervals.
result Improves robust accuracy, compressing the gap between theory and practice.
Paper proposes efficient algorithms to certify robustness of ReLU networks.
problem NP-complete problem of verifying robustness for ReLU networks.
method Two computationally efficient algorithms Fast-Lin and Fast-Lip.
result Delivers bounds close to exact minimum distortion with significant speedup.
Paper improves neural network robustness certification with tighter radii estimates.
problem Certifying neural networks' robustness against adversarial attacks.
method Advanced algorithms for discrete and continuous domains, optimizing sample size, standard deviation, and temperature.
result Significant improvement in certified test-set accuracy with tighter certified radii bounds.
Paper tightens optimization bounds using conformal prediction.
problem Lack of practical informiveness in dual bounds from optimization solvers.
method Introduces conformal prediction framework to tighten loose primal and dual bounds.
result Proposed method produces tighter, more informative prediction intervals.
Improved defense against data poisoning attacks by aggregating smaller subsets.
problem Mitigating the impact of poisoned data on model robustness.
method Finite Aggregation method that combines duplicates of smaller disjoint subsets for training.
result Consistent improvement in certified robustness bounds, up to 4.77% on GTSRB.
New method certifies deep graph classifiers with tighter risk bounds.
problem Certifying the reliability of deep graph classifiers.
method Linearized deep assignment flows with random initial conditions, using PAC-Bayes risk certification.
result Computes tighter out-of-sample risk certificates efficiently.
New framework provides robustness guarantees against adversarial attacks.
problem Adversarial examples lead to different outputs from deep-learning algorithms.
method Connects robustness to additive noise and proposes a training strategy.
result Scalable method improves certified bounds on adversarial perturbation.
Unified framework certifies predictor performance under distribution shift.
problem Certifying predictor performance under distribution shift.
method Unified framework with explicit inequalities, sound verification, and identifiable structure.
result Explicit upper bound on excess risk under shift.
New method defends against patch attacks with high-certainty guarantees.
problem Patch attacks on images, especially physical adversarial attacks.
method Randomized smoothing, exploiting patch constraints.
result Meaningfully large robustness certificates against patch attacks.
Paper improves robustness certification by integrating ML and logical reasoning.
problem Limited robustness certification under perturbation radius.
method Integrates statistical ML models with logical reasoning using Markov logic networks.
result First certified robustness bound for MLN derived and experimentally validated.
Improves robustness of GNNs with minimal loss in accuracy.
problem Non-robustness of GNNs to adversarial attacks on node attributes.
method Certifiable robustness method for binary node attributes and L_0-bounded perturbations, combined with robust semi-supervised training.
result Certified robustness and non-robustness of GNNs, with minimal loss in accuracy.
First certified defense scaling to large datasets and models.
problem Robustness against adversarial examples in machine learning models.
method PixelDP, based on differential privacy.
result First certified defense that scales to large networks and models.
Certified calibration methods protect model confidence from adversarial attacks.
problem Adversarial attacks degrade model calibration, reducing confidence in predictions.
method Developed certified calibration methods to provide worst-case bounds on calibration under adversarial perturbations.
result Certified calibration methods produce analytic and approximate bounds for the Brier score and expected calibration error.
This paper provides a comprehensive benchmark and taxonomy for certifiably robust DNN defenses.
problem Certifiably robust defenses against adversarial attacks for deep neural networks.
method Taxonomy and benchmark of certifiably robust approaches.
result First comprehensive benchmark of certifiably robust approaches on different datasets.
Paper certifies intersection of minimum-volume confidence sets for multinomial outcomes.
problem Certifying intersection of minimum-volume confidence sets for multinomial outcomes.
method Exploits likelihood ordering to induce halfspace constraints, enabling adaptive geometric partitioning and computable bounds on p-values.
result Efficient and provably sound algorithm for certifying intersection, disjointness, or indeterminate result.
Automates perturbation analysis for neural networks, enabling certified robustness on complex architectures.
problem Limited applicability of existing perturbation analysis methods to complex neural network architectures.
method Developed an automatic framework to generalize LiRPA algorithms to any neural network structure, enabling loss fusion and state-of-the-art certified defense results.
result Demonstrated LiRPA based certified defense on Tiny ImageNet and Downscaled ImageNet.
Paper provides efficient robustness certificates for neural networks.
problem Ensuring neural networks are robust against adversarial attacks.
method Two-step approach: 1) Efficient convex optimization for robustness certificates with bounded Hessian eigenvalues, 2) Curvature-based regularization during training.
result Significantly higher certified robust accuracy achieved compared to existing methods.
New semidefinite relaxation improves robustness certification of neural networks.
problem Certifying robustness of neural networks against adversarial examples.
method Proposed a new semidefinite relaxation for certifying robustness of arbitrary ReLU networks.
result Our proposed relaxation is tighter than previous relaxations and produces meaningful robustness guarantees.
New method makes machine learning models robust to label flipping attacks.
problem Machine learning models are vulnerable to label flipping attacks.
method Randomized smoothing over arbitrary functions to build certifiably robust classifiers.
result Linear classifiers are robust to label flipping attacks with deterministic bounds.
This work improves adversarial robustness in sparse coding models.
problem The gap between theoretical models and practical deployment in adversarial robustness.
method Combining a sparsity-promoting encoder with a linear classifier, and providing a robustness certificate.
result Bounding the robust risk and providing a robustness certificate for end-to-end classification.
Deep Partition Aggregation defends against poisoning attacks with provable certificates.
problem Adversarial poisoning attacks corrupt classifier test-time behavior.
method Deep Partition Aggregation (DPA) is an ensemble method using hash partitions and base models trained on these partitions.
result DPA can certify >= 50% of test images against over 500 poison image insertions on MNIST, and nine insertions on CIFAR-10.
High-dimensional smoothing techniques struggle with robustness guarantees against various attacks.
problem Challenges in extending randomized smoothing to other attack models in high-dimensional space.
method Analysis of isotropic and generalized Gaussian smoothing distributions, proving bounds on certified robustness radii.
result Certifiable robustness radii decrease as $O(1/d^{ rac{1}{2} - rac{1}{p}})$ with dimension d for p>2.