Go to NNW-NNT AI Modules

Function
Ref. Model
I/O Data
SubAIMs
JSON MData
Profiles
Ref. Software
Conformance
Performance

1 Functions

The Watermark Extraction (NNW-WEX) AIM provides the Retrieved payload in response to the Modified model of a neural network, that is, the model of a watermarked neural network that may have been modified after the insertion. Given the Locator, it looks where the locator says; without it, where its method looks.

2 Reference Model

3 Input/Output Data

Table 1 specifies the Input and Output Data of the Watermark Extraction (NNW-WEX) AIM.Table 1 – I/O Data of the Watermark Extraction (NNW-WEX)

Input Acronyms Descriptions
Modified model AIP-MLO The neural network model, possibly modified after the insertion, from which the Payload is to be retrieved.
Locator NNW-WLC Where the watermark sits in the model and the method that inserted it, as the insertion AIM gave it. Optional: without it the AIM uses what its method defaults to.
Output Acronyms Descriptions
Retrieved payload NNW-PLD The information retrieved from the model.

4 SubAIMs

No SubAIMs.

5 JSON Metadata

https://schemas.mpai.community/NNW1/V1.2/AIMs/WatermarkExtraction.json

6 Profiles

No profiles.

7 Reference Software

8 Conformance Testing

9 Performance Assessment

Go to NNW-NNT AI Modules