How Robust Is Image Classification Deep Neural Network Watermarking?

Explore the robustness of image classification deep neural network watermarking in this 17-minute IEEE conference talk. Delve into the motivations, contributions, and key concepts surrounding watermarking techniques for neural networks. Examine the watermarking taxonomy, attack taxonomy, decision threshold, and Nash equilibrium. Analyze the setup, runtime, and fidelity after embedding, comparing single schemes against all attacks. Gain valuable insights into guidelines and conclusions drawn from this systematic study on the effectiveness of deep neural network watermarking methods.

Intro
Motivation
Contributions
Watermarking Definition
Watermarking Taxonomy
Attack Taxonomy
Decision Threshold
Nash Equilibrium
Setup
Runtime Analysis
Fidelity after Embedding
Single Scheme vs All Attacks
Discussion
Guidelines
Conclusion