Influences and Boolean functions representations
Seminar Room 1, Newton Institute
More than twenty years ago the important work of Kahn, Kalai and Linial gave general bounds on the influence of variables in arbitrary Boolean functions over the discrete hypercube. In theoretical computer science, though, often one is interested in particular types of "simple" Boolean functions such as constant-depth circuits, decision trees, low-degree polynomial threshold functions, etc. This additional structure raises the possibility that refined influence bounds can be obtained, and indeed it turns out that this is sometimes the case. This talk will give an overview of several such results and their applications, with an emphasis on currently open questions and directions for future work.