Go To AIP Data Types

1 Definition 2 Functional Requirements 3 Syntax 4 Semantics 5    Conformance Testing 6     Performance Assessment

1 Definition

A Data Type conveying information about the function, the training and the Data interfaces of a Machine Learning Model.

2 Functional Requirements

The ML Model Manifest includes:

  1. The ID of the M-Instance and the ID of the ML Model Manifest.
  2. The name and the version of the ML Model.
  3. The function performed by the ML Model, expressed in natural language.
  4. The IDs of the training datasets used to train the ML Model.
  5. The Application Profile of the ML Model.
  6. The Data Type that the ML Model takes as input, with its acronym and name.
  7. The Data Type that the ML Model produces as output, with its acronym and name.
  8. Data Exchange Metadata providing additional information on the ML Model Manifest.

3 Syntax

https://schemas.mpai.community/AIP/V1.0/data/MLModelManifest.json

4 Semantics

Label Description
Header ML Model Manifest Header – Standard “AIP-MLM-Vx.y”
MInstanceID Identifier of M-Instance.
MLModelManifestID Identifier of the ML Model Manifest.
MLModelManifestTime Time of the ML Model Manifest.
MLModelName Name of the ML Model.
ModelVersion Version of the ML Model.
MLModelFunction The function of the ML Model, described in natural language.
TrainingDataSetIDs The IDs of the training datasets used to train the ML Model.
ApplicationProfile The Application Profile of the ML Model.
– ProfileType Type of the Application Profile.
MLModelInputDataTypes The Data Type that the ML Model takes as input.
– DataType The Data Type.
– – Acronym Acronym of the Data Type, in the form XXX-YYY-Vn.m.
– – Name Name of the Data Type.
MLModelOutputDataTypes The Data Type that the ML Model produces as output.
– DataType The Data Type.
– – Acronym Acronym of the Data Type, in the form XXX-YYY-Vn.m.
– – Name Name of the Data Type.
DataXMData Information about this ML Model Manifest instance.
DescrMetadata Descriptive Metadata.

5 Conformance Testing

A Data instance Conforms with ML Model Manifest (AIP-MLM) if:

  1. Its JSON Object validates against its JSON Schema.
  2. Any included JSON Object validates against its JSON Schema.
  3. All Data in the JSON Object:
    1. Have the specified Data Types.
    2. Conform with the Qualifiers signaled in their JSON Schemas.

6 Performance Assessment

Go To AIP Data Types