: Reconstructing sensitive training data from a model's predictions to compromise privacy. Deep Learning for Defense
Deep learning is increasingly used to build more robust security systems: Collection of Deep Learning Cyber Security Research Papers security
: Injecting malicious data into training sets to corrupt the learning process. : Reconstructing sensitive training data from a model's
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 security
: Reverse-engineering a trained model to reveal its parameters or architecture.