Relieve Adversarial Attacks Based on Multimodal Training
This paper explores the role of multimodal training in mitigating the problems caused by adversarial attacks, building on the foundations of deep learning. Deep learning models have reached great success in many areas such as image recognition and natural language processing. But their robustness ha...
Saved in:
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
EDP Sciences
2025-01-01
|
| Series: | ITM Web of Conferences |
| Online Access: | https://www.itm-conferences.org/articles/itmconf/pdf/2025/01/itmconf_dai2024_02004.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1825206539792678912 |
|---|---|
| author | Lai Hongjie |
| author_facet | Lai Hongjie |
| author_sort | Lai Hongjie |
| collection | DOAJ |
| description | This paper explores the role of multimodal training in mitigating the problems caused by adversarial attacks, building on the foundations of deep learning. Deep learning models have reached great success in many areas such as image recognition and natural language processing. But their robustness has always been a concern. However, the emergence of adversarial attacks has exposed shortages of neural networks, forcing people to confront their limitations and further increasing concerns about the security of deep learning models. Adversarial training is an effective defense mechanism that incorporates adversarial samples into the training data, enabling models to better detect and resist attacks. This paper first introduces the principles and types of adversarial attacks, as well as basic concepts and related methods, including Fast Gradient Sign Method (FGSM), Projected Gradient Descent (PGD), DeepFool, and Jacobian Saliency Map Attack (JSMA). The paper then focuses on analyzing the robustness of the multimodal model CLIP based on contrastive learning. Finally, the paper proposes whether audio data can be added to the training samples of the CLIP model to further improve its robustness, and raises related issues and bottlenecks. |
| format | Article |
| id | doaj-art-0595d09c27a84055bdec205c3e623526 |
| institution | Kabale University |
| issn | 2271-2097 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | EDP Sciences |
| record_format | Article |
| series | ITM Web of Conferences |
| spelling | doaj-art-0595d09c27a84055bdec205c3e6235262025-02-07T08:21:10ZengEDP SciencesITM Web of Conferences2271-20972025-01-01700200410.1051/itmconf/20257002004itmconf_dai2024_02004Relieve Adversarial Attacks Based on Multimodal TrainingLai Hongjie0Sydney Smart Technology College, Northeastern University at QinhuangdaoThis paper explores the role of multimodal training in mitigating the problems caused by adversarial attacks, building on the foundations of deep learning. Deep learning models have reached great success in many areas such as image recognition and natural language processing. But their robustness has always been a concern. However, the emergence of adversarial attacks has exposed shortages of neural networks, forcing people to confront their limitations and further increasing concerns about the security of deep learning models. Adversarial training is an effective defense mechanism that incorporates adversarial samples into the training data, enabling models to better detect and resist attacks. This paper first introduces the principles and types of adversarial attacks, as well as basic concepts and related methods, including Fast Gradient Sign Method (FGSM), Projected Gradient Descent (PGD), DeepFool, and Jacobian Saliency Map Attack (JSMA). The paper then focuses on analyzing the robustness of the multimodal model CLIP based on contrastive learning. Finally, the paper proposes whether audio data can be added to the training samples of the CLIP model to further improve its robustness, and raises related issues and bottlenecks.https://www.itm-conferences.org/articles/itmconf/pdf/2025/01/itmconf_dai2024_02004.pdf |
| spellingShingle | Lai Hongjie Relieve Adversarial Attacks Based on Multimodal Training ITM Web of Conferences |
| title | Relieve Adversarial Attacks Based on Multimodal Training |
| title_full | Relieve Adversarial Attacks Based on Multimodal Training |
| title_fullStr | Relieve Adversarial Attacks Based on Multimodal Training |
| title_full_unstemmed | Relieve Adversarial Attacks Based on Multimodal Training |
| title_short | Relieve Adversarial Attacks Based on Multimodal Training |
| title_sort | relieve adversarial attacks based on multimodal training |
| url | https://www.itm-conferences.org/articles/itmconf/pdf/2025/01/itmconf_dai2024_02004.pdf |
| work_keys_str_mv | AT laihongjie relieveadversarialattacksbasedonmultimodaltraining |