5 years ago

Category: Security

There are five major security loopholes in TensorFlow models, such as programs, untrusted models and input, and other vulnerabilities.