1 Definition 2 Functional Requirements 3 Syntax 4 Semantics

1      Definition

Machine Learning Model Qualifier is a Data Type providing additional information on Machine Learning Model Data for potential use by a machine.

The combination of Machine Learning Model Data and Machine Learning Model Qualifier is called Machine Learning Model Object. Machine Learning Model is specified by MMM-AIF V2.2.

2      Functional Requirements

Machine Learning Model Qualifier has the following functional requirements:

  1. Sub-Type
    1. MLModel Types
    2. NNModel Types
  2. Format
    1. Extension
    2. Framework
    3. Exchange
  3. Attributes
    1. Regulation
    2. Certification Type
    3. Validity

Users needing additional entries in the3D Model Qualifier or support of new Qualifiers should make a documented request to the MPAI Secretariat. Requests will be considered by the appropriate MPAI committee.

3     Syntax

https://schemas.mpai.community/TFA/V1.4/data/MLModelQualifier.json

4       Semantics

  1. Sub-Type

    1. Bayesian Classifier
    2. Neural Network
      1. Convolutional with residuals
      2. Convolutional without residuals
      3. Deconvolutional
      4. Eco-State Neural Network
      5. Feed Forward
      6. Modular
      7. Recurrent
      8. Transformer
    3. Random Forest
    4. Support Vector Machine
    5. XG-Boost

  2. Format

    1. Extension
      1. Engine
      2. H5
      3. Keras
      4. Mlmodel
      5. Onnx
      6. Pd
      7. Pkl
      8. Pt
      9. Pth
      10. Tflite
    2. Framework
      1. ai
      2. Caffe
      3. CNTK
      4. Keras
      5. MxNet
      6. PyTorch
      7. Scikit-Learn
      8. TensorFlow
    3. Exchange
      1. ONNX
      2. NNEF

  3. Attributes

    1. Regulations
      1. AI Act
      2. CAIO
      3. Data Act
      4. Data Governance Act
      5. EBSI
      6. GDPR
    2. Certification Type
      1. Bylaw
      2. Sandbox
    3. Validity (Time)