Nonlinear Dimensionality Reduction Frameworks

22
Nonlinear Dimensionality Reduction Frameworks Rong Xu Chan su Lee

description

Nonlinear Dimensionality Reduction Frameworks. Rong Xu Chan su Lee. Outline. Intuition of Nonlinear Dimensionality Reduction(NLDR) ISOMAP LLE NLDR in Gait Analysis. Intuition: how does your brain store these pictures?. Brain Representation. Brain Representation. Every pixel? - PowerPoint PPT Presentation

Transcript of Nonlinear Dimensionality Reduction Frameworks

Page 1: Nonlinear Dimensionality Reduction Frameworks

Nonlinear Dimensionality Reduction Frameworks

Rong Xu

Chan su Lee

Page 2: Nonlinear Dimensionality Reduction Frameworks

Outline

• Intuition of Nonlinear Dimensionality Reduction(NLDR)

• ISOMAP

• LLE

• NLDR in Gait Analysis

Page 3: Nonlinear Dimensionality Reduction Frameworks

Intuition: how does your brain store these pictures?

Page 4: Nonlinear Dimensionality Reduction Frameworks

Brain Representation

Page 5: Nonlinear Dimensionality Reduction Frameworks

Brain Representation

• Every pixel?

• Or perceptually meaningful structure?– Up-down pose

– Left-right pose

– Lighting direction

So, your brain successfully reduced the high-dimensional inputs to an intrinsically 3-dimensional manifold!

Page 6: Nonlinear Dimensionality Reduction Frameworks

Data for Faces

Page 7: Nonlinear Dimensionality Reduction Frameworks

Data for Handwritten “2”

Page 8: Nonlinear Dimensionality Reduction Frameworks

Data for Hands

Page 9: Nonlinear Dimensionality Reduction Frameworks

Manifold Learning

• A manifold is a topological space which is locally Euclidean

• An example of nonlinear manifold:

Page 10: Nonlinear Dimensionality Reduction Frameworks

Manifold Learning• Discover low dimensional

representations (smooth manifold) for data in high dimension.

• Linear approaches(PCA, MDS) vs Non-linear approaches (ISOMAP, LLE)

d

i Ry Y

XN

i Rx

latent

observed

Page 11: Nonlinear Dimensionality Reduction Frameworks

Linear Approach- PCA

• PCA Finds subspace linear projections of input data.

Page 12: Nonlinear Dimensionality Reduction Frameworks

Linear Approach- MDS

• MDS takes a matrix of pairwise distances and gives a mapping to Rd. It finds an embedding that preserves the interpoint distances, equivalent to PCA when those distance are Euclidean.

• BUT! PCA and MDS both fail to do embedding with nonlinear data, like swiss roll.

Page 13: Nonlinear Dimensionality Reduction Frameworks

Nonlinear Approaches- ISOMAP

• Constructing neighbourhood graph G• For each pair of points in G, Computing shortest

path distances ---- geodesic distances.• Use Classical MDS with geodesic distances.

Euclidean distance Geodesic distance

Josh. Tenenbaum, Vin de Silva, John langford 2000

Page 14: Nonlinear Dimensionality Reduction Frameworks

Sample points with Swiss Roll

• Altogether there are 20,000 points in the “Swiss roll” data set. We sample 1000 out of 20,000.

Page 15: Nonlinear Dimensionality Reduction Frameworks

Construct neighborhood graph G

K- nearest neighborhood (K=7)

DG is 1000 by 1000 (Euclidean) distance matrix of two neighbors (figure A)

Page 16: Nonlinear Dimensionality Reduction Frameworks

Compute all-points shortest path in G

Now DG is 1000 by 1000 geodesic distance matrix of two arbitrary points along the manifold(figure B)

Page 17: Nonlinear Dimensionality Reduction Frameworks

Find a d-dimensional Euclidean space Y(Figure c) to minimize the cost function:

Use MDS to embed graph in Rd

Page 18: Nonlinear Dimensionality Reduction Frameworks

22

222

22

2

2

)(min||||min

||)(5.0||min||)()(||min

1*11

2

1)(

2

1)(

YYXXtraceYYXX

HDDHDD

NIH

DD

HHDD

HHDD

TTTT

GYG

ijG

YY

GG

Y

Theorem: For any squared distance matrix ,there exists of points xi and,xj, such that

So

2GD

)()'(2

jijiij xxxxd

XXD TG 2

Linear Approach-classical MDS

Page 19: Nonlinear Dimensionality Reduction Frameworks

Solution

dxNdddxN

N

v

v

v

yyyY

22

11

21

|

|

|

|

|

|

VY dY2/1

)(

Y lies in Rd and consists of N points correspondent to the N original points in input space.

Page 20: Nonlinear Dimensionality Reduction Frameworks

PCA, MD vs ISOMAP

Page 21: Nonlinear Dimensionality Reduction Frameworks

• Nonlinear

• Globally optimal• Still produces globally optimal low-dimensional Euclidean

representation even though input space is highly folded,

twisted, or curved.

• Guarantee asymptotically to recover the true dimensionality.

Isomap: Advantages

Page 22: Nonlinear Dimensionality Reduction Frameworks

• May not be stable, dependent on topology of data

• Guaranteed asymptotically to recover geometric structure of nonlinear manifolds– As N increases, pairwise distances provide better

approximations to geodesics, but cost more computation

– If N is small, geodesic distances will be very inaccurate.

Isomap: Disadvantages