Potential cyber threats of adversarial attacks on autonomous driving models

Boltachev, Eldar

doi:10.1007/s11416-023-00486-x

Cited by 6 publications

(1 citation statement)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…• Wenkai Tan(2023) [22] • Eldar Boltachev(2023) [4] discussed the scientific literature and highlighted the main types of disruptive attacks on autonomous driving models that pose potential threats to CAVs. In this paper, they compiled a dataset with traffic sign images obtained from public sources and made experiments in which we distorted the original images and used them to train deep neural network-based classification models.…”

Section: Literature Review 1 Overviewmentioning

confidence: 99%

Defense Artificial Neural Networks From Adversarial Attacks by Using Deep Learning Classification Techniques

2024

IRJMETS

View full text Add to dashboard Cite

In the recent years, Artificial neural networks become a significant part of a per-son's everyday life. It is a computational model inspired by the human's brain neural structure which consists of interconnected nodes and information flows through these nodes and the network adjusts the connection strengths during training to learn from data, recognize patterns, make predictions and solve var-ious tasks in DL/ML and artificial intelligence. It contributes to the creation of a smarter world by connecting intelligence to things or entities that use the network and integrating them into multiple fields that offer helpful services. But the lack of transparency raises concerns, especially in critical applications like healthcare , AVs systems , security areas, or finance. In security area one of the most common attack is the adversarial attacks which exploit the model's sensitivity to minor changes/perturbations that pose a critical challenge to the robustness of machine learning models. These attacks increasing day by day. Classification ML/DL techniques are utilized for the defences of these type of attacks. A detailed review and comparative analysis of existing ML/DL techniques for defences of attacks on ANNs has been done. Also a complete review and comparative analysis of existing work done in defences of Adversarial attacks on ANNs. The comparative analysis of various techniques helps us for deciding which approach is much suitable for defences purpose of Adversarial attacks on ANNs.

show abstract

Section: Literature Review 1 Overviewmentioning

confidence: 99%