Estimating entropy of distributions in constant space
We consider the task of estimating the entropy of k-ary distributions from samples in the streaming model, where space is limited. Our main contribution is an algorithm that requires O ( klog(1"3/")2 ) samples and a constant O(1) memory words of space and outputs a ±" estimate of H(p)...
Main Author: | |
---|---|
Other Authors: | |
Format: | Article |
Language: | English |
Published: |
Neural Information Processing Systems Foundation,
2021-01-25T19:53:49Z.
|
Subjects: | |
Online Access: | Get fulltext |