Impact Score 6.76
Deep learning has been widely used in a vast number of applications, such as natural language processing, computer vision, decision making, data security, etc. To learn effective deep learning models, it requires a lot of time and efforts for data collection and computation power allocation. These models may be illegally distributed for unauthorized usages, which sacrifices the interest of the model owners. On the other hand, the deep learning models are also vulnerable to adversarial samples or poisoned training data. This severely reduces the accuracy and reliability of deep learning-based techniques. The research of deep learning model security is to ensure the authenticity and reliability of deep learning models against various attacks. It can be mainly categorized into the area of 1) deep learning model attacks, such as the adversarial sample generation, data poisoning, training data reconstruction, and 2) deep learning model defense, such as model watermarking, adversarial sample detection, adversarial training. The research on deep learning model security is still in its infancy, though some progress have been made, it is far from satisfactory towards training a robust and reliable model for deep learning-based applications.
This special issue aims at promoting the research on both the attacks and defense on deep learning models, which includes fast adversarial sample generation schemes, imperceptible data poisoning schemes, accurate training data reconstruction schemes, robust and traceable watermarking schemes for deep learning models, strategies for defending against the adversarial samples. Related researchers and engineers can contribute with original research that present their work. All submitted papers will be peer-reviewed and selected on the basis of both their quality and relevance to the theme of this special issue. Topics of interest for this special issue include, but are not limited to:
Topics of interest
Manuscript submission deadline: June 20, 2022
Final acceptance deadline: January 31, 2023
Prospective authors are invited to upload their papers through the Editorial Manager system. The platform will start accepting submissions approximately one week before the submission period begins. When submitting your manuscript please select the article type “VSI:DLMS”. It is mandatory that the manuscripts be submitted by the deadline indicated. All manuscripts should adhere to the Journal’s guidelines: please take into account that Special Issue papers follow the same submission rules as regular articles.
All submissions deemed suitable to be sent for peer review will be evaluated by at least two independent reviewers. Once your manuscript is accepted, it will go into production, and will be simultaneously published in the current regular issue and pulled into the online Special Issue. Articles from this Special Issue will appear in different regular issues of the journal, though they will be clearly marked and branded as Special Issue articles.
Please see an example here: https://www.sciencedirect.com/journal/science-of-the-total-environment/special-issue/10SWS2W7VVV
Please make sure you have read the Guide for Authors before preparing your manuscript. The Guide for Authors and the link to submit your manuscript are available on the Journal’s homepage.