Download presentation
1
Nearest Neighbor Searching Under Uncertainty
Wuzhou Zhang Supervised by Pankaj K. Agarwal Department of Computer Science Duke University
2
Nearest Neighbor Searching (NNS)
Applications Pattern Recognition, Data Compression Statistical Classification, Clustering Databases, Information Retrieval Computer Vision, etc.
3
Nearest Neighbor Searching Under Uncertainty
Discrete pdf Continuous pdf
4
Nearest Neighbor In Expectation
_________
5
Bisector In Case Of Gaussian
For Gaussian distribution, bisector is a line! Hard to get explicit formula! Figure:
6
Squared Distance Function
bisector is simple and beautiful! In case of discrete pdf, bisector is also a line! In both cases, compute the Voronoi diagram, solve it optimally! However, not a metric !
7
Sampling Continuous Distributions
Sometimes working on continuous distributions is hard…. Lower bounds on other metrics and distributions are also possible…. Let’s focus on discrete pdf then….
8
Expected Nearest Neighbor In L1 Metric (Manhattan metric)
9
Expected Nearest Neighbor In L1 Metric ( cont. )
Source: Range Searching on Uncertain Data [P.K.Agarwal et al. 2009]
10
Geometric Reduction
11
Building Block: Half-Space Intersection and Convex Hulls
Upper hulls correspond to lower envelopes, an example in 2D Source: page 252 – 253, Computational Geometry: Algorithms and Applications, 3rd Edition[Mark de Berg et al. ]
12
Segment-tree Based Data Structures for Expected-NN In L1 Metric
13
Segment-tree Based Data Structures for Expected-NN In L1 Metric ( cont
14
Segment-tree Based Data Structures for Expected-NN In L1 Metric ( cont
Size of data structure Preprocessing time Query time Summary of the result
15
Approximate L2 Metric It’s a metric when P is centrally symmetric!
16
Approximate L2 Metric ( cont. )
More complex!
17
Future Work Approximate the expected NN in L2 metric
Work harder in the near future! Approximate the expected NN in L2 metric Study the complexity of expected Voronoi diagram Study the probability case
18
Questions? Thanks! Main References:
[1] Pankaj K. Agarwal, Siu-Wing Cheng, Yufei Tao, Ke Yi: Indexing uncertain data. PODS 2009: [2] Pankaj K. Agarwal, Lars Arge, Jeff Erickson: Indexing Moving Points. J. Comput. Syst. Sci. 66(1): (2003) Questions?
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.