Security Threats from Data Poisoning Attacks in Deep Learning Vision Systems
Open AccessThe practice of using deep learning methods in safety-critical visionsystems such as autonomous driving has come a long way. As vision systems supported by deep learning methods become ubiquitous, the possible security threats faced by these systems have come into greater focus. As with any artificial intelligence system, these deep neural vision networks are first trained on a data set of interest; once they start performing well, they are deployed to a real-world environment. In the training stage, deep learning systems are susceptible to data poisoning attacks.Although deep neural networks have proved to be versatile in solving ahost of challenges, these systems have complex data ecosystems, especially in computer vision. In practice, the security threats that are encountered when training these systems are often ignored when deploying these models in the real world. However, these threats pose significant risks to the overall reliability of the system.This research presents the fundamentals of data poisoning attackswhen training deep learning vision systems and discusses countermeasures against these types of attacks. In addition, it simulates the risk posedby a real-world data poisoning-attack on a deep learning vision system and presents a novel algorithm, MOVCE - Model Verification with Convolutional Neural Network and Word Embeddings - which provides an effective countermeasure for maintaining the reliability of the system. The countermeasure described in this research can be used in a wide variety of use cases where the risks posed by poisoning training data are similar.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.
| Thumbnail | Title | Date Uploaded | Visibility | Actions |
|---|---|---|---|---|
|
|
Raghavan_gwu_0075A_16293.pdf | 2023-11-14 | Open Access |
|