Links:
LSH Primer: https://github.com/FALCONN-LIB/FALCONN/wiki/LSH-Primer
Github LSHash: https://github.com/kayzhu/LSHash
图像检索:再叙ANN Search:https://yongyuan.name/blog/ann-search.html
Practical and optimal LSH for angular distance: http://papers.nips.cc/paper/5893-practical-and-optimal-lsh-for-angular-distance.pdf
如果本身数据分布有一定特殊性(如集中在某一区域),标准分布生成的平面可能会产生比较差的切分,个人感觉还要去掉这些比较差的切分.