1.1 - 8.10 - IDWT2D (ML Engine) - Teradata Vantage

Teradata Vantage™ - Machine Learning Engine Analytic Function Reference

Teradata Vantage
Release Number
October 2019
Content Type
Programming Reference
Publication ID
English (United States)

The IDWT2D function is the inverse of DWT2D (ML Engine); that is, IDWT2D applies inverse wavelet transforms on multiple sequences simultaneously. IDWT2D takes as input the output table and meta table output by DWT2D and outputs the sequences as 2-dimensional matrixes. (Because the IDWT2D output is comparable to the DWT2D input, the inverse transformation is also called the reconstruction.)

How Machine Learning Engine function IDWT2D works

This is a typical IDWT2D use case:

  1. Apply DWT2D to 2-dimensional sequences to create the coefficients of the matrixes and corresponding metadata.
  2. Filter the coefficients by methods appropriate for the objects (for example, minimum threshold or top n coefficients), compressing the original matrixes.
  3. Apply IDWT to the filtered coefficients to reconstruct the sequences.
  4. Compare the reconstructed matrixes to their original counterparts.