Tuesday, June 25, 2024

SWARM – Switchable Backdoor Attack Against Pre-trained Models

In the big data era, pre-training large vision transformer (ViT) models on massive datasets has become prevalent for enhanced performance on downstream tasks. 

Visual prompting (VP), introducing learnable task-specific parameters while freezing the pre-trained backbone, offers an efficient adaptation alternative to full fine-tuning. 

However, the VP’s potential security risks remain unexplored. The following cybersecurity analysts from Tsinghua University, Tencent Security Platform Department, Zhejiang University, Research Center of Artificial Intelligence, Peng Cheng Laboratory recently uncovered a novel backdoor attack threat for VP in a cloud service scenario, where a threat actors can attach or remove an extra “switch” prompt token to toggle between clean and backdoored modes stealthily:-

  • Sheng Yang
  • Jiawang Bai
  • Kuofeng Gao
  • Yong Yang

SWARM – Switchable Backdoor Attack

Researchers’ proposed Switchable Attack against pre-trained Models (SWARM) optimizes a trigger, clean prompts, and the switch token via clean loss, backdoor loss, and cross-mode feature distillation, ensuring normal behavior without the switch while forcing target misclassification when activated.

ANYRUN malware sandbox’s 8th Birthday Special Offer: Grab 6 Months of Free Service

Experiments across visual tasks demonstrate SWARM’s high attack success rate and evasiveness.

Here an offending cloud service provider acts as a threat actor, this is based on existing backdoor attack scenarios.

These users submit task datasets and pre-trained models to the threat actor’s service. 

They also apply the trained API of attackers while attempting to identify and mitigate backdoors.

The opponent does not manage user samples but controls prompt inputs. In normal mode, a model should handle triggered patterns without any detection

In backdoor mode, it should have a high attack success rate. This attack aims at hiding triggers by predicting correctly on clean samples and misclassifying them when a “switch” trigger is added. 

Clean and Backdoor mode (Source – Arxiv)

The threat actor understands the downstream dataset and tunes prompts accordingly through visual prompting.

Visual prompting adds learnable prompt tokens after the embedding layer so that during training only these task-specific parameters are modified.

Users may use augmented clean data and mitigation techniques such as Neural Attention Distillation (NAD) and I-BAU to address this risk.

While, the researchers’ experiments reveal that SWARM achieves 96% ASR against NAD and over 97% against I-BAU, as a result outperforming baseline attacks by a significant margin. 

This shows SWARM’s ability to evade detection and mitigate threats, which consequently increases the danger to victims. 

Researchers propose a new brand of backdoor attack on adapting pre-trained vision transformers with visual prompts, which insert an extra switch token for making invisible transitions between clean mode and backdoored one. 

SWARM indicates a new realm of attack mechanisms while also providing acceleration for future defense research.

Free Webinar on Live API Attack Simulation: Book Your Seat | Start protecting your APIs from hackers


Latest articles

Hackers Attacking Windows IIS Server to Upload Web Shells

Windows IIS Servers often host critical web applications and services that provide a gateway...

WikiLeaks Founder Julian Assange Released in Stunning Deal with U.S.

WikiLeaks founder Julian Assange has been released from prison after reaching a deal with...

Four Members of FIN9 Hackers Charged for Attacking U.S. Companies

Four Vietnamese nationals have been charged for their involvement in a series of computer...

BREAKING: NHS England’s Synnovis Hit by Massive Cyber Attack

In a shocking development, the NHS has revealed that it was the victim of...

Threat Actor Claiming a 0-day in Linux LPE Via GRUB bootloader

A new threat actor has emerged, claiming a zero-day vulnerability in the Linux GRUB...

LockBit Ransomware Group Claims Hack of US Federal Reserve

The notorious LockBit ransomware group has claimed responsibility for hacking the U.S. Federal Reserve,...

Microsoft Power BI Vulnerability Let Attackers Access Organizations Sensitive Data

A vulnerability in Microsoft Power BI allows unauthorized users to access sensitive data underlying...
Tushar Subhra Dutta
Tushar Subhra Dutta
Tushar is a Cyber security content editor with a passion for creating captivating and informative content. With years of experience under his belt in Cyber Security, he is covering Cyber Security News, technology and other news.

Free Webinar

API Vulnerability Scanning

71% of the internet traffic comes from APIs so APIs have become soft targets for hackers.Securing APIs is a simple workflow provided you find API specific vulnerabilities and protect them.In the upcoming webinar, join Vivek Gopalan, VP of Products at Indusface as he takes you through the fundamentals of API vulnerability scanning..
Key takeaways include:

  • Scan API endpoints for OWASP API Top 10 vulnerabilities
  • Perform API penetration testing for business logic vulnerabilities
  • Prioritize the most critical vulnerabilities with AcuRisQ
  • Workflow automation for this entire process

Related Articles