IRLS using DL frameworks

• IRLS(Iterative re-weighted least square) for Logistic Regression, implemented using

  • tensorflow < 2.0
  • tensorflow2.0
  • pytorch
  • megengine

• Note that IRLS is a second order optimization problem, which is equivalent to Newton's method.

• We show that these DL frameworks can do general matrix based algorithms, and can be accelerated by the power of gpu.

• In this implementation, we use svd to solve pseudo inverse of singular matrices.