Robust Mean Estimation on Highly Incomplete Data with Arbitrary Outliers
We study the problem of robustly estimating the mean of a $d$-dimensional distribution given $N$ examples, where most coordinates of every example may be missing and $\varepsilon N$ examples may be arbitrarily corrupted. Assuming each coordinate appears in a constant factor more than $\varepsilon N$...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Published: |
2020
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2008.08071 http://arxiv.org/abs/2008.08071 |
| Tags: |
No Tags, Be the first to tag this record!
|