DriveGuard: Robustification of Automated Driving Systems with Deep Spatio-Temporal Convolutional Autoencoder

Papachristodoulou, Andreas; Kyrkou, Christos; Theocharides, Theocharis

doi:10.1109/wacvw52041.2021.00016

Cited by 4 publications

(2 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…If V ′ p differs from V p significantly, the input node p is more likely an anomalous node. NODLINK numerically measures the difference between V ′ p and V p with the normalized MSELoss, following other VAE-based anomaly detectors [108], [81]. To follow the common terms in machine learning, we use the reconstruction error RE to represent the value of the MSELoss between V ′ p and V p .…”

Section: B Terminal Identificationmentioning

confidence: 99%

NODLINK: An Online System for Fine-Grained APT Attack Detection and Investigation

Li,

Dong,

Xiao

et al. 2024

Proceedings 2024 Network and Distributed System Security Symposium

View full text Add to dashboard Cite

Section: B Terminal Identificationmentioning

confidence: 99%

NODLINK: An Online System for Fine-Grained APT Attack Detection and Investigation

Li,

Dong,

Xiao

et al. 2024

Proceedings 2024 Network and Distributed System Security Symposium

View full text Add to dashboard Cite

“…However, their focus is not robustness against perturbed images as only clean images are used in training. DriveGuard [25] explores different autoencoder architectures on adversarially degraded images that affect semantic segmentation rather than the steering task. They show that autoencoders can be used to enhance the quality of the degraded images, thus improving overall task performance.…”

Section: Related Workmentioning

confidence: 99%

AutoJoin: Efficient Adversarial Training for Robust Maneuvering via Denoising Autoencoder and Joint Learning

Villarreal¹,

Poudel²,

Ryan³

et al. 2022

Preprint

View full text Add to dashboard Cite

As a result of increasingly adopted machine learning algorithms and ubiquitous sensors, many 'perception-to-control' systems have been deployed in various settings. For these systems to be trustworthy, we need to improve their robustness with adversarial training being one approach. In this work, we propose a gradientfree adversarial training technique, called AutoJoin. AutoJoin is a very simple yet effective and efficient approach to produce robust models for imaged-based autonomous maneuvering. Compared to other SOTA methods with testing on over 5M perturbed and clean images, AutoJoin achieves significant performance increases up to the 40% range under perturbed datasets while improving on clean performance for almost every dataset tested. In particular, AutoJoin can triple the clean performance improvement compared to the SOTA work by Shen et al. [31]. Regarding efficiency, AutoJoin demonstrates strong advantages over other SOTA techniques by saving up to 83% time per training epoch and 90% training data. The core idea of AutoJoin is to use a decoder attachment to the original regression model creating a denoising autoencoder within the architecture. This allows the tasks 'steering' and 'denoising sensor input' to be jointly learnt and enable the two tasks to reinforce each other's performance.

show abstract

AutoJoin: Efficient Adversarial Training against Gradient-Free Perturbations for Robust Maneuvering via Denoising Autoencoder and Joint Learning

Villarreal,

Poudel,

Wickman

et al. 2024

2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

View full text Add to dashboard Cite

DriveGuard: Robustification of Automated Driving Systems with Deep Spatio-Temporal Convolutional Autoencoder

Cited by 4 publications

References 32 publications

NODLINK: An Online System for Fine-Grained APT Attack Detection and Investigation

NODLINK: An Online System for Fine-Grained APT Attack Detection and Investigation

AutoJoin: Efficient Adversarial Training for Robust Maneuvering via Denoising Autoencoder and Joint Learning

AutoJoin: Efficient Adversarial Training against Gradient-Free Perturbations for Robust Maneuvering via Denoising Autoencoder and Joint Learning

Contact Info

Product

Resources

About