Algebraic techniques for constructing minimal weight threshold functions

Abstract

A linear threshold element computes a function that is a sign of a weighted sum of the input variables. The best known lower bounds on the size of threshold circuits are for depth-2 circuits with small (polynomial-size) weights. However, in general, the weights are arbitrary integers and can be of exponential size in the number of input variables. Namely, obtaining progress in lower bounds for threshold circuits seems to be related to understanding the role of large weights. In the present literature, a distinction is made between the two extreme cases of linear threshold functions with polynomial-size weights, as opposed to those with exponential-size weights. Our main contributions are in devising two novel methods for constructing threshold functions with minimal weights and filling up the gap between polynomial and exponential weight growth by further refining the separation. Namely, we prove that the class of linear threshold functions with polynomial-size weights can be divided into subclasses according to the degree of the polynomial. In fact, we prove a more general result — that there exists a minimal weight linear threshold function for any arbitrary number of inputs and any weight size.