The intersection of security and deep learning covers two primary areas: using deep learning to security (e.g., intrusion detection) and protecting deep learning models from vulnerabilities (e.g., adversarial attacks) . Key Security Threats to Deep Learning
Researchers focus on several critical vulnerabilities that can compromise AI models:
: Injecting malicious data into training sets to corrupt the learning process.
: Subtly altering input data to trick a model into making incorrect predictions.