From the course: CompTIA SecAI+ (CY0-001) Cert Prep

Unlock this course with a free trial

Join today to access over 25,600 courses taught by industry experts.

Transfer learning attacks

Transfer learning attacks

From the course: CompTIA SecAI+ (CY0-001) Cert Prep

Start my 1-month free trial Buy for my team

Transfer learning attacks

“

Transfer learning allows developers to start with a pre-trained model and fine-tune it on a smaller domain-specific dataset. Transfer learning provides a powerful shortcut. Instead of training from scratch, you build on an existing foundation. This approach helps when data or compute resources are limited. That convenience also creates risk because attackers can target the components and workflows that support transfer learning. When you adopt a pre-trained model, you also inherit everything inside of it. You inherit the training biases, the embedded behaviors, and any hidden vulnerabilities or malicious artifacts. If an attacker compromises the original model, those issues can persist into fine-tuned versions. They can become harder to spot because fine-tuning can mask or distort the original model's behavior. Attackers exploit this by distributing tainted models through public repositories, by using legitimate sounding names, and by embedding these models in tutorials that look…

Contents