Knowledge Graph — Coursera Notes › Academic disciplines › Computer Science / Information Technology › Artificial Intelligence › Responsible AI

Adversarial attacks

concept · part of Responsible AI

Small input manipulations trick the model into wrong predictions.

Inside Adversarial attacks (2)

• Data poisoning — An attacker injects false data into the training set, making the model learn wrong patterns.
• Model inversion — An attacker reverse-engineers a model to reconstruct its training data.

Connections

• Related to Data poisoning
• Uses Adversarial robustness
• Related to Adversarial robustness

This is the text view of an interactive 3D knowledge graph — open this page with JavaScript enabled to explore it visually.