Absolute error for selected values of N, ℓ and k; step quantization (SQ) approach.
dataset
posted on 2015-03-26, 03:00 authored by José Portêlo, Bhiksha Raj, Isabel TrancosoN corresponds to the number of inputs for the logsum, ℓ corresponds to the number of bits representing the inputs and k is the number of entries of the look-up table for the piecewise linear approximation. In each cell, the first and second rows contain the mean and maximum absolute error with respect to the true value of the logsum, respectively.
Absolute error for selected values of N, ℓ and k; step quantization (SQ) approach.
