The paper is devoted to show that there are simple and accurate ways to compute a sigmoid nonlinearity in digital h ardware by piece-wise linearization. This is done as the computations involved are complex, but even more interesting, for the data compression performed by the sigmoid, which can half the number of Boolean functions to be further implemented. We detail such an approximation algorithm, and analyze its ac curacy, showing im- provements over the previous known algorithms. The algorithm uses only ad - ditions/subtractions and shifts (multiplications by powers of 2), supporting our claim of "simplicity".
|Title of host publication||European Symposium on Artificial Neural Networks|
|Publication status||Published - Apr 20 1994|
|Event||ESANN'94 - Brussels, Belgium|
Duration: Apr 20 1994 → …
|Period||4/20/94 → …|