Security and Adversarial AI
Now that we have our first end-to-end AI service, we will discuss how to secure it and demonstrate why traditional cybersecurity is inadequate against the new breed of adversarial AI attacks. We will perform our first adversarial attack on our image recognition service. We will define adversarial AI and discuss how it relates to conventional security problems.
More specifically, you will learn about fundamental security concepts, effective approaches to secure computer systems, and how to apply baseline security to our adversarial AI playground and how this relates to adversarial AI. This will help you do the following:
- Understand fundamental security concepts such as the CIA triad, security frameworks such as NIST, threat modeling, security controls, DevSecOps, and MLOps, and how they all relate
- Secure and harden your deployment host
- Secure your environment from intruders, restrict access to the Service API, and protect against Denial...
