加载中…
Let $ E={ (1, b_1, \ldots , b_n)\in R^{n+1} \mid ; b_i= \pm 1 ,; i=1, \ldots, n }E^{\times n}{\ne 0} := { W=(w{i_1}, \ldots , w_{i_n}) \mid w_{i_k}\in E, , k=1, \ldots, n, , dim , span(w_{i_1}, \ldots , w_{i_n}) = n },q^W_l := |span(w_{i_{n-l+1}}, \ldots , w_{i_n}) \cap E|.$ Then for any weights , , we have for the number of threshold functions the following lower bound and the right side of the inequality doesn’t depend on the choice of . Here the indices used in the numerator correspond to vectors from $span(w_{i_1}, \ldots , w_{i_n})\cap E = \left{w_{i_1}, \ldots, w_{i_n}, \ldots w_{i_{q_n^W}}\right}$.

生成讲解博客的中文版本

生成AI智能校验的中文版本