Search papers, labs, and topics across Lattice.
This paper explores the relationships and separations among three measures of binary concept classes: sign rank, $\mathbb{Z}_2$-index, and list replicability number. By establishing that the $\mathbb{Z}_2$-index is upper-bounded by a linear function of the list replicability number, the authors provide a significant separation between sign rank and $\mathbb{Z}_2$-index, addressing a previously unresolved question. Additionally, the study introduces upper bounds on the list replicability number and a composition result that enhances the understanding of these measures in learning theory.
A strong separation between sign rank and $\mathbb{Z}_2$-index reveals new insights into the complexity of binary concept classes.
In learning theory, the sign rank of a binary concept class captures the smallest dimension in which it can be represented by points and halfspaces. Despite tremendous interest, lower bounds on sign rank are notoriously difficult to come by. Two recent approaches to the problem establish lower bounds on sign rank by measures that are easier to analyze: the $\mathbb{Z}_2$-index and the list replicability number. We order these measures, showing that the $\mathbb{Z}_2$-index is upper-bounded by a linear function of the list replicability number. As a main consequence, we obtain a strong separation between sign rank and $\mathbb{Z}_2$-index, thereby resolving a question of Frick, Hosseini, and Vasileuski. This motivates a thorough study of list replicability, the stronger of the two lower-bounding measures. We establish upper bounds on the list replicability number by two combinatorial measures: height and minimum star number. We also prove a fundamental composition result, showing that the product of two concept classes has list replicability number bounded by the sum of the list replicability numbers of the two classes.
