Subquadratic Kronecker regression with applications to tensor decomposition (NeurIPS 2022)

subquadratic-kronecker-regression

Files

• kronecker_regression_main.py -- Implementation of FastKroneckerRegression and DJSSW19.
• run_all_kronecker_regression.py -- Wrapper to generate Figure 1 and Table 1.
• tucker_als.py -- Library with KronMatMul and TuckerALS implementations.
• tucker_decomposition_experiments.py -- Main file for running Tucker decomposition with command line arguments.
• run_all_tensor_decomposition.py -- Wrapper to run all Tucker decomposition experiments.
• tensor_data_handler.py -- Handler class to read different tensor data into np.ndarray format.
• data/ -- Contains instructions on how to generate all of our datasets.
• output/ -- Contains the output logs for all of our Kronecker regression and tensor decomposition experiments.
• plots/ -- Contains code to generate Figure 1.

Instructions

• $ python3 run_all_kronecker_regression.py
• $ python3 run_all_tensor_decomposition.py

Source: https://github.com/fahrbach/subquadratic-kronecker-regression