We tackle some fundamental problems in probability theory on corrupted random
processes on the integer line. We analyze when a biased random walk is expected
to reach its bottommost point and when intervals of integer points can be
detected under a natural model of noise. We apply these results to problems in
learning thresholds and intervals under a new model for learning under
adversarial design.
Description
[2003.13561] On Biased Random Walks, Corrupted Intervals, and Learning Under Adversarial Design
%0 Journal Article
%1 berend2020biased
%A Berend, Daniel
%A Kontorovich, Aryeh
%A Reyzin, Lev
%A Robinson, Thomas
%D 2020
%K adversarial randomized readings robustness sampling
%T On Biased Random Walks, Corrupted Intervals, and Learning Under
Adversarial Design
%U http://arxiv.org/abs/2003.13561
%X We tackle some fundamental problems in probability theory on corrupted random
processes on the integer line. We analyze when a biased random walk is expected
to reach its bottommost point and when intervals of integer points can be
detected under a natural model of noise. We apply these results to problems in
learning thresholds and intervals under a new model for learning under
adversarial design.
@article{berend2020biased,
abstract = {We tackle some fundamental problems in probability theory on corrupted random
processes on the integer line. We analyze when a biased random walk is expected
to reach its bottommost point and when intervals of integer points can be
detected under a natural model of noise. We apply these results to problems in
learning thresholds and intervals under a new model for learning under
adversarial design.},
added-at = {2020-04-01T20:26:40.000+0200},
author = {Berend, Daniel and Kontorovich, Aryeh and Reyzin, Lev and Robinson, Thomas},
biburl = {https://www.bibsonomy.org/bibtex/21c2503a5422695d8a83471710bc219e5/kirk86},
description = {[2003.13561] On Biased Random Walks, Corrupted Intervals, and Learning Under Adversarial Design},
interhash = {6e05e2770e800e3f134ebb2144230c81},
intrahash = {1c2503a5422695d8a83471710bc219e5},
keywords = {adversarial randomized readings robustness sampling},
note = {cite arxiv:2003.13561Comment: 18 pages},
timestamp = {2020-04-01T20:26:40.000+0200},
title = {On Biased Random Walks, Corrupted Intervals, and Learning Under
Adversarial Design},
url = {http://arxiv.org/abs/2003.13561},
year = 2020
}