Gemoetrically local embedding in manifolds for dimension reduction

20
Intelligent Database Systems Presenter : Kung, Chien-Hao Authors : Shuzhi Sam Ge, Hongsheng He, Chengyao Shen 2012,PR Gemoetrically local embedding in manifolds for dimension reduction

description

Gemoetrically local embedding in manifolds for dimension reduction. Presenter : Kung, Chien-Hao Authors : Shuzhi Sam Ge , Hongsheng He, Chengyao Shen 2012,PR. Outlines. Motivation Objectives Methodology Experiments Conclusions Comments. Motivation. - PowerPoint PPT Presentation

Transcript of Gemoetrically local embedding in manifolds for dimension reduction

Page 1: Gemoetrically  local embedding in  manifolds for dimension reduction

Intelligent Database Systems Lab

Presenter : Kung, Chien-Hao

Authors : Shuzhi Sam Ge, Hongsheng He, Chengyao Shen

2012,PR

Gemoetrically local embedding in manifolds for dimension reduction

Page 2: Gemoetrically  local embedding in  manifolds for dimension reduction

Intelligent Database Systems Lab

Outlines

MotivationObjectivesMethodologyExperimentsConclusionsComments

Page 3: Gemoetrically  local embedding in  manifolds for dimension reduction

Intelligent Database Systems Lab

Motivation• LLE is a dimension reduction

technique which preserve

neighborhood relationships amongst

data.

• However, Euclidean distance is

limited as only the pairwise distance

to the target data is considered.

Page 4: Gemoetrically  local embedding in  manifolds for dimension reduction

Intelligent Database Systems Lab

Objectives• This paper uses geometry distance which emphasized

the local geometrical structure of the manifold

spanned instead of computing the pairwise metric

between data.

Page 5: Gemoetrically  local embedding in  manifolds for dimension reduction

Intelligent Database Systems Lab

Methodology-FrameworkGeometrical distance

construction

Optimal reconstruction

Outlier-suppressingembedding

Page 6: Gemoetrically  local embedding in  manifolds for dimension reduction

Intelligent Database Systems Lab

MethodologyNeighbor selection using geometry distances

Tikhonov regularization

Page 7: Gemoetrically  local embedding in  manifolds for dimension reduction

Intelligent Database Systems Lab

MethodologyAlternative neighbor selection

Page 8: Gemoetrically  local embedding in  manifolds for dimension reduction

Intelligent Database Systems Lab

MethodologyLinear embedding

Page 9: Gemoetrically  local embedding in  manifolds for dimension reduction

Intelligent Database Systems Lab

MethodologyOutlier data filtering

Page 10: Gemoetrically  local embedding in  manifolds for dimension reduction

Intelligent Database Systems Lab

Experiment

Page 11: Gemoetrically  local embedding in  manifolds for dimension reduction

Intelligent Database Systems Lab

Experiment

Page 12: Gemoetrically  local embedding in  manifolds for dimension reduction

Intelligent Database Systems Lab

Experiment

Page 13: Gemoetrically  local embedding in  manifolds for dimension reduction

Intelligent Database Systems Lab

Experiment

Page 14: Gemoetrically  local embedding in  manifolds for dimension reduction

Intelligent Database Systems Lab

Experiment

Page 15: Gemoetrically  local embedding in  manifolds for dimension reduction

Intelligent Database Systems Lab

Experiment

Page 16: Gemoetrically  local embedding in  manifolds for dimension reduction

Intelligent Database Systems Lab

Experiment

Page 17: Gemoetrically  local embedding in  manifolds for dimension reduction

Intelligent Database Systems Lab

Experiment

Page 18: Gemoetrically  local embedding in  manifolds for dimension reduction

Intelligent Database Systems Lab

Experiment

Page 19: Gemoetrically  local embedding in  manifolds for dimension reduction

Intelligent Database Systems Lab

Conclusions• The GLE algorithm performs well in extracting inner

structures of input linear manifold with outliers.

• The GLE behaves as a clustering and classification method by projecting the feature data into low-dimensional separable regions.

• The major drawback of GLE is the slow computation speed compared with other algorithms when the input data is small.

Page 20: Gemoetrically  local embedding in  manifolds for dimension reduction

Intelligent Database Systems Lab

Comments• Advantages– This paper supplies the completely formula

information. But this paper is hard to understand when the reader is a lack of prior knowledge.

• Applications– Dimension reduction.