Foreword

Introduction

Scope

Definitions

References

 Use Cases

Robustness Evaluation

Computational Cost Evaluation

Imperceptibility Evaluation
Technical Specification: Neural Network Watermarking (MPAI-NNW) V1.0 specifies methods to evaluate the following aspects of Active (Watermarking) and Passive (Fingerprinting) Neural Network Traceability Methods:

  • Ability of a Neural Network Traceability Detector/Decoder to detect/decode/match Traceability Data when the traced Neural Network has been modified.
  • Computational cost of injecting, extracting, detecting, decoding, or matching Traceability Data.
  • Specifically for active tracing methods, impact on the performance of a neural network with inserted Traceability Data and its inference.

The MPAI-NNT V1.0 Technical Specification is published with a request for Community Comments. Anybody can submit comments to the MPAI Secretariat until 2024/12/12T 23:59 UTC. Comments will be considered when drafting the final text of MPAI-NNT V1.0. The current version of the Technical Specification will be presented online on 2024/12/10T15 UTC. Register to participate.

WARNING

Use of the technologies described in this Technical Specification may infringe patents, copyrights or intellectual property rights of MPAI Members or non-members.

MPAI and its Members accept no responsibility whatsoever for damages or liability, direct or consequential, which may result from the use of this Technical Specification.

Readers are invited to review Notices and Disclaimers.

© Copyright MPAI 2021-24.  All rights reserved.