Table of Contents

DimensionalityReduction

This stereotype was considered for one concrete case in the Brandeis program and has become deprecated.

The idea was to consider dimensionality reduction of vector spaces as a privacy mechanism.

Required input objects

2 data objects (data, projection matrix)

Required output objects

1 data object (feature vector)

Restrictions

Restriction: task must have exactly two input and one output data objects
Error: exactly 2 inputs and 1 output are required

Restriction: “data” property must exist under stereotype details object - “data” must have been selected and saved and “data” information must not have been deleted from model’s file
Error: data is undefined

Restriction: “projectionMatrix” property must exist under stereotype details object - “projectionMatrix” must have been selected and saved and “projectionMatrix” information must not have been deleted from model’s file
Error: projectionMatrix is undefined

Restriction: input data object fixed (by its id) in “data” property must exist also on model
Error: data object is missing

Restriction: input projectionMatrix object fixed (by its id) in “projectionMatrix” property must exist also on model
Error: projectionMatrix object is missing

Restriction: data and projectionMatrix objects must be different data objects - they must have different ids
Error: data and projectionMatrix must be different objects