Affiliation:
1. Princeton University, Princeton, NJ
2. Sandia National Labs, Livermore, Livermore, CA
Abstract
We investigate a new class of geometric problems based on the idea of online error correction. Suppose one is given access to a large geometric dataset though a query mechanism; for example, the dataset could be a terrain and a query might ask for the coordinates of a particular vertex or for the edges incident to it. Suppose, in addition, that the dataset satisfies some known structural property
P
(for example, monotonicity or convexity) but that, because of errors and noise, the queries occasionally provide answers that violate
P
. Can one design a filter that modifies the query's answers so that (i) the output satisfies
P
; (ii) the amount of data modification is minimized? We provide upper and lower bounds on the complexity of online reconstruction for convexity in 2D and 3D.
Funder
National Science Foundation
Army Research Office
Publisher
Association for Computing Machinery (ACM)
Subject
Artificial Intelligence,Hardware and Architecture,Information Systems,Control and Systems Engineering,Software
Cited by
5 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献