Algorithms for Verifying Deep Neural Networks

Liu, Changliu; Arnon, Tomer; Lazarus, Christopher; Strong, Christopher A.; Barrett, Clark; Kochenderfer, Mykel J.

doi:10.48550/arxiv.1903.06758

Cited by 48 publications

(58 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Such works are complementary to ours in the sense that they provide a convergence analysis of an existing algorithm for deep learning. In a different line of work, Liu et al (2019a) propose to exploit interpolation to prove convergence of a new acceleration method for deep learning. However, their experiments suggest that the method still requires the use of a hand-designed learning-rate schedule.…”

Section: Related Workmentioning

confidence: 99%

A Stochastic Bundle Method for Interpolating Networks

Paren¹,

Berrada²,

Poudel³

et al. 2022

Preprint

View full text Add to dashboard Cite

We propose a novel method for training deep neural networks that are capable of interpolation, that is, driving the empirical loss to zero. At each iteration, our method constructs a stochastic approximation of the learning objective. The approximation, known as a bundle, is a pointwise maximum of linear functions. Our bundle contains a constant function that lower bounds the empirical loss. This enables us to compute an automatic adaptive learning rate, thereby providing an accurate solution. In addition, our bundle includes linear approximations computed at the current iterate and other linear estimates of the DNN parameters. The use of these additional approximations makes our method significantly more robust to its hyperparameters. Based on its desirable empirical properties, we term our method Bundle Optimisation for Robust and Accurate Training (BORAT). In order to operationalise BORAT, we design a novel algorithm for optimising the bundle approximation efficiently at each iteration. We establish the theoretical convergence of BORAT in both convex and non-convex settings. Using standard publicly available data sets, we provide a thorough comparison of BORAT to other single hyperparameter optimisation algorithms. Our experiments demonstrate BORAT matches the state-of-the-art generalisation performance for these methods and is the most robust.

show abstract

Section: Related Workmentioning

confidence: 99%

A Stochastic Bundle Method for Interpolating Networks

Paren¹,

Berrada²,

Poudel³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…This makes it difficult to quantify how the output varies given controlled semantically-aligned variations in the input from our unit tests. While there are works that aim to bring interpretable formal verification to DL models [9,10], the scale is still far from the millions if not billions of parameters used in contemporary models [11][12][13].…”

Section: Introductionmentioning

confidence: 99%

Auditing AI models for Verified Deployment under Semantic Specifications

Bharadhwaj¹,

Huang²,

Xiao³

et al. 2021

Preprint

View full text Add to dashboard Cite

Auditing trained deep learning (DL) models prior to deployment is vital in preventing unintended consequences. One of the biggest challenges in auditing is in understanding how we can obtain human-interpretable specifications that are directly useful to the end-user. We address this challenge through a sequence of semantically-aligned unit tests, where each unit test verifies whether a predefined specification (e.g., accuracy over 95%) is satisfied with respect to controlled and semantically aligned variations in the input space (e.g., in face recognition, the angle relative to the camera). We perform these unit tests by directly verifying the semantically aligned variations in an interpretable latent space of a generative model. Our framework, AuditAI, bridges the gap between interpretable formal verification and scalability. With evaluations on four different datasets, covering images of towers, chest X-rays, human faces, and ImageNet classes, we show how AuditAI allows us to obtain controlled variations for verification and certified training while addressing the limitations of verifying using only pixel-space perturbations. A blog post accompanying the paper is at this link auditai.github.io

show abstract

“…Another approach to address safety for RL policies is to utilize a safety layer which filters out unsafe actions as proposed in [1], [13], [14]. This safety layer must be easily verifiable without any black boxes such as deep neural networks inside its architecture which are hard to verify [15]. In [1] authors used safety constraints in order to prevent unsafe lane changes from a RL agent.…”

Section: Introductionmentioning

confidence: 99%

Minimizing Safety Interference for Safe and Comfortable Automated Driving with Distributional Reinforcement Learning

Kamran¹,

Engelgeh²,

Busch³

et al. 2021

Preprint

View full text Add to dashboard Cite

Despite recent advances in reinforcement learning (RL), its application in safety critical domains like autonomous vehicles is still challenging. Although punishing RL agents for risky situations can help to learn safe policies, it may also lead to highly conservative behavior. In this paper, we propose a distributional RL framework in order to learn adaptive policies that can tune their level of conservativity at run-time based on the desired comfort and utility. Using a proactive safety verification approach, the proposed framework can guarantee that actions generated from RL are fail-safe according to the worst-case assumptions. Concurrently, the policy is encouraged to minimize safety interference and generate more comfortable behavior. We trained and evaluated the proposed approach and baseline policies using a high level simulator with a variety of randomized scenarios including several corner cases which rarely happen in reality but are very crucial. In light of our experiments, the behavior of policies learned using distributional RL can be adaptive at run-time and robust to the environment uncertainty. Quantitatively, the learned distributional RL agent drives in average 8 seconds faster than the normal DQN policy and requires 83% less safety interference compared to the rule-based policy with slightly increasing the average crossing time. We also study sensitivity of the learned policy in environments with higher perception noise and show that our algorithm learns policies that can still drive reliable when the perception noise is two times higher than the training configuration for automated merging and crossing at occluded intersections.

show abstract

Algorithms for Verifying Deep Neural Networks

Cited by 48 publications

References 33 publications

A Stochastic Bundle Method for Interpolating Networks

A Stochastic Bundle Method for Interpolating Networks

Auditing AI models for Verified Deployment under Semantic Specifications

Minimizing Safety Interference for Safe and Comfortable Automated Driving with Distributional Reinforcement Learning

Contact Info

Product

Resources

About