1 Introduction (Informative)
Traceability enables identification of the origin or verification of the integrity of data. In the case of Neural Networks, Traceability enables understanding and tracking the development, use, and evolution of Neural Network models. A variety of methods have been developed for NN Traceability, and can be divided into two categories:
- Watermarking method, an Active method which alters the Weights of the NN to insert Traceability Data.
- Fingerprinting method, a Passive method which does not alter the Weights of the NN.
Numerous Traceability methods have been published since 2017 [6], especially for watermarking.
Technical Specification: Neural Network Watermarking (MPAI-NNW) V1.0 provides tools to evaluate Watermarking methods, for a given Payload, on three properties: Imperceptibility, Robustness, and Computational Cost.
Technical Specification: Neural Network Traceability (MPAI-NNT) V1.0 provides tools to evaluate both types of Traceability methods keeping the methods included in MPAI-NNW V1.0.
In all Chapters and Sections, Terms beginning with a capital letter are defined in Table 1 if they are specific to this Technical Specification and in Table 2 if they are common to all MPAI Technical Specifications. All Chapters, Sections, and Annexes are Normative unless they are labelled as Informative.
