6
Motivations and main challenge Experiments A Relative Dependency Test Wacha BOUNLIPHONE, Arthur GRETTON, Arthur TENENHAUS, Matthew BLASCHKO Equipe GALEN 1/5

A Relative Dependency Test - Inria · Motivations and main challenge Experiments A Relative Dependency Test Wacha BOUNLIPHONE, Arthur GRETTON, Arthur TENENHAUS, Matthew BLASCHKO Equipe

  • Upload
    dodien

  • View
    216

  • Download
    0

Embed Size (px)

Citation preview

Motivations and main challenge Experiments

A Relative Dependency Test

Wacha BOUNLIPHONE, Arthur GRETTON, ArthurTENENHAUS, Matthew BLASCHKO

Equipe GALEN

1 / 5

Motivations and main challenge Experiments

Motivations

2 / 5

Motivations and main challenge Experiments

Motivations

3 / 5

Motivations and main challenge Experiments

Main challenge

Main Challenge

To develop new statistical method to determine whether adependency is stronger between one pair of measurements oranother.

Dependent(Source,Target1) > Dependent(Source,Target2) ?

4 / 5

Motivations and main challenge Experiments

Verification and calibration

Illustration of synthetic datasets

In this dataset, we can control the relative degree of functionaldependency between variates.

−1 0 1 2 3 4 5 6 7−2

−1.5

−1

−0.5

0

0.5

1

1.5

2

−10 −5 0 5 10 15−15

−10

−5

0

5

10

−15 −10 −5 0 5 10 15−15

−10

−5

0

5

10

(a) (b) (c)

Question: DependentTest(a,b) > DependentTest(a,c) ?

p-v

alu

e

0 0.2 0.4 0.6 0.8 10

0.2

0.4

0.6

0.8

1

p−values

freq

uenc

y

0 0.2 0.4 0.6 0.8 10

0.2

0.4

0.6

0.8

1

p−values

freq

uenc

y

γ3 Type I error Type II error5 / 5

Motivations and main challenge Experiments

Pediatric high-grade gliomas (pHGG)

Brain tumors localisation

pHGG have different genetics origins depending on the location ofthe tumor in the brain. The goal is to identify the mechanismsresponsible of the tumor.

0 1 2 3 4 5 6

x 10−3

−2

−1

0

1

2

3

4

5

6x 10−3

HSIC1XY

HS

IC1X

Z

dependent testindependent test

6 / 5