Optimal Transport vs. Fisher-Rao distance between Copulas

Gautier Marti
Gautier MartiAI Quant at Shell Street Labs
Introduction
Statistical distances
Optimal Transport vs. Fisher-Rao distance
between Copulas
IEEE SSP 2016
G. Marti, S. Andler, F. Nielsen, P. Donnat
June 28, 2016
Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
Introduction
Statistical distances
Clustering of Time Series
We need a distance Dij between time series xi and xj
If we look for ‘correlation’, Dij is a decreasing function of ρij ,
a measure of ‘correlation’
Several choices are available for ρij . . .
Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
Introduction
Statistical distances
Copulas
Sklar’s Theorem:
F(xi , xj ) = Cij (Fi (xi ), Fj (xj ))
Cij , the copula, encodes the dependence structure
Fr´echet-Hoeffding bounds:
max{ui + uj − 1, 0} ≤ Cij (ui , uj ) ≤ min{ui , uj }
(left) lower-bound, (mid) independence, (right) upper-bound copulas
Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
Introduction
Statistical distances
Copulas - Gaussian Example
Gaussian copula: CGauss
R (ui , uj ) = ΦR(Φ−1(ui ), Φ−1(uj ))
The distribution is parametrized by a correlation matrix R.
Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
Introduction
Statistical distances
The Target/Forget (copula-based) Dependence Coefficient
Dependence is measured as the relative distance from independence to
the nearest target-dependence: comonotonicity or counter-monotonicity
Which distances are appropriate between copulas for the task of
clustering (copulas and time series)?
Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
Introduction
Statistical distances
Definitions - Fisher-Rao geodesic distance
Metrization of the paramater space {θ ∈ Rd | p(X; θ)dx = 1}.
Consider the metric gjk(θ) = − ∂2 log p(x,θ)
∂θj ∂θk
p(x, θ)dx,
the infinitesimal length ds(θ) = ( θ) G(θ) θ,
the Fisher-Rao geodesic distance
FR(θ1, θ2) =
θ2
θ1
ds(θ).
f -divergences induce infinitesimal length proportional to
Fisher-Rao infinitesimal length:
Df (θ θ + dθ) =
1
2
( θ) G(θ) θ.
Thus, they have the same local behaviour [1].
Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
Introduction
Statistical distances
Definitions - Optimal Transport distances
Wasserstein metric
Wp(µ, ν)p
= inf
γ∈Γ(µ,ν) M×M
d(x, y)p
dγ(x, y)
Image from Optimal Transport for Image Processing, Papadakis
Other transportation distances: regularized discrete optimal
transport [3], Sinkhorn distances [2], . . .
Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
Introduction
Statistical distances
Geometry of covariances
Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
Introduction
Statistical distances
Distances between Gaussian copulas
Copulas C1, C2, C3 encoding a correlation of 0.5, 0.99, 0.9999 respectively;
Which pair of copulas is the nearest?
- For Fisher-Rao, Kullback-Leibler, Hellinger and related divergences:
D(C1, C2) ≤ D(C2, C3);
- For Wasserstein: W2(C2, C3) ≤ W2(C1, C2)
Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
Introduction
Statistical distances
Distances as a function of (ρ1, ρ2)
Distance heatmap and surface as a function of (ρ1, ρ2)
for Fisher-Rao for Wasserstein W2
Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
Introduction
Statistical distances
Distances impact on clustering
Datasets of bivariate time series are generated from six Gaussian copulas
with correlation .1, .2, .6, .7, .99, .9999
Distance heatmaps for Fisher-Rao (left), W2 (right); Using Ward
clustering, Fisher-Rao yields clusters of copulas with correlations
{.1, .2, .6, .7}, {.99}, {.9999}, W2 yields {.1, .2}, {.6, .7}, {.99, .9999}
Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
Introduction
Statistical distances
Fisher metric and the Cram´er–Rao lower bound
Cram´er–Rao lower bound (CRLB)
The variance of any unbiased estimator ˆθ of θ is bounded by the
reciprocal of the Fisher information G(θ):
var(ˆθ) ≥
1
G(θ)
.
In the bivariate Gaussian copula case,
var(ˆρ) ≥
(ρ − 1)2(ρ + 1)2
3(ρ2 + 1)
.
Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
Introduction
Statistical distances
Fisher metric and the Cram´er–Rao lower bound
We consider the set of 2 × 2 correlation matrices C =
1 θ
θ 1
parameterized by θ.
Let x =
x1
x2
∈ R2
.
f (x; θ) = 1
2π 1−θ2
exp − 1
2
x C−1
x = 1
2π 1−θ2
exp − 1
2(1−θ2)
(x2
1 + x2
2 − 2θx1x2)
log f (x; θ) = − log(2π 1 − θ2) − 1
2(1−θ2)
(x2
1 + x2
2 − 2θx1x2)
∂2 log f (x;θ)
∂θ2 = − θ2+1
(θ2−1)2 −
x2
1
2(θ+1)3 +
x2
1
2(θ−1)3 −
x2
2
2(θ+1)3 +
x2
2
2(θ−1)3 −
x1x2
(θ+1)3 −
x1x2
(θ−1)3
Then, we compute ∞
−∞
∂2 log f (x;θ)
∂θ2 f (x; θ)dx.
Since E[x1] = E[x2] = 0, E[x1x2] = θ, E[x2
1 ] = E[x2
2 ] = 1, we get
∞
−∞
∂2 log f (x;θ)
∂θ2 f (x; θ)dx =
− θ2+1
(θ2−1)2 − 1
2(θ+1)3 + 1
2(θ−1)3 − 1
2(θ+1)3 + 1
2(θ−1)3 − θ
(θ+1)3 − θ
(θ−1)3 = −
3(θ2+1)
(θ−1)2(θ+1)2
Thus,
G(θ) =
3(θ2
+ 1)
(θ − 1)2(θ + 1)2
.
Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
Introduction
Statistical distances
Fisher metric and the Cram´er–Rao lower bound
In the bivariate Gaussian copula case,
var(ˆρ) ≥
(ρ − 1)2(ρ + 1)2
3(ρ2 + 1)
.
Recall that locally Fisher-Rao and the f -divergences are a
quadratic form of the Fisher metric ( θ) G(θ) θ. So, the
discriminative power of these distances is well calibrated with
respect to statistical uncertainty. For this purpose, they induce the
appropriate curvature on the parameter space.
Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
Introduction
Statistical distances
Properties of these distances
In addition, for clustering we prefer OT since:
in a parametric setting:
Fisher-Rao and f -divergences are defined on density manifolds,
but some important copulas (such as the Fr´echet-Hoeffding
upper bound) do not belong to these manifolds;
Thus, in case of closed-form formulas (such as in the Gaussian
case), they are ill-defined for these copulas (for perfect
dependence, covariance is not invertible)
in a non-parametric/empirical setting:
f -divergences are defined for absolutely continuous measures,
thus require a pre-processing KDE
they are not aware of the support geometry, thus badly handle
noise on the support
Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
Introduction
Statistical distances
Barycenters
OT is defined for both discrete/empirical and continuous measures
and is support-geometry aware:
0 0.5 1
0
0.5
1
0.0000
0.0015
0.0030
0.0045
0.0060
0.0075
0.0090
0.0105
0.0120
0 0.5 1
0
0.5
1
0.0000
0.0015
0.0030
0.0045
0.0060
0.0075
0.0090
0.0105
0.0120
0 0.5 1
0
0.5
1
0.0000
0.0008
0.0016
0.0024
0.0032
0.0040
0.0048
0.0056
0 0.5 1
0
0.5
1
0.0000
0.0015
0.0030
0.0045
0.0060
0.0075
0.0090
0.0105
0.0120
0 0.5 1
0
0.5
1
0.0000
0.0015
0.0030
0.0045
0.0060
0.0075
0.0090
0.0105
0.0120
5 copulas describing the dependence between X ∼ U([0, 1]) and
Y ∼ (X ± i )2
, where i is a constant noise specific for each distribution
0 0.5 1
0
0.5
1 Wasserstein barycenter copula
0.0000
0.0004
0.0008
0.0012
0.0016
0.0020
0.0024
0.0028
0.0032
Barycenter of the 5 copulas for a divergence and OT
Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
Introduction
Statistical distances
Future Research
Develop further geometries of copulas
using Optimal Transport: show that dependence-clustering of
time series is improved over standard correlations
using f -divergences: detect efficiently dependence-regime
switching in multivariate time series (cf. Fr´ed´eric Barbaresco’s
work on radar signal processing)
Numerical experiments and code:
https://www.datagrapple.com/Tech/fisher-vs-ot.html
Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
Introduction
Statistical distances
Shun-ichi Amari and Andrzej Cichocki.
Information geometry of divergence functions.
Bulletin of the Polish Academy of Sciences: Technical
Sciences, 58(1):183–195, 2010.
Marco Cuturi.
Sinkhorn distances: Lightspeed computation of optimal
transport.
In Advances in Neural Information Processing Systems, pages
2292–2300, 2013.
Sira Ferradans, Nicolas Papadakis, Julien Rabin, Gabriel Peyr´e,
and Jean-Fran¸cois Aujol.
Regularized discrete optimal transport.
Springer, 2013.
Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
1 of 18

Recommended

InfoGAIL by
InfoGAIL InfoGAIL
InfoGAIL Sungjoon Choi
2.7K views55 slides
Building Biomedical Knowledge Graphs for In-Silico Drug Discovery by
Building Biomedical Knowledge Graphs for In-Silico Drug DiscoveryBuilding Biomedical Knowledge Graphs for In-Silico Drug Discovery
Building Biomedical Knowledge Graphs for In-Silico Drug DiscoveryVaticle
60 views136 slides
Support vector regression and its application in trading by
Support vector regression and its application in tradingSupport vector regression and its application in trading
Support vector regression and its application in tradingAashay Harlalka
2.5K views12 slides
Rabin Carp String Matching algorithm by
Rabin Carp String Matching  algorithmRabin Carp String Matching  algorithm
Rabin Carp String Matching algorithmsabiya sabiya
2.3K views16 slides
Deep Learning by
Deep LearningDeep Learning
Deep LearningJun Wang
5.2K views111 slides
cCorrGAN: Conditional Correlation GAN for Learning Empirical Conditional Dist... by
cCorrGAN: Conditional Correlation GAN for Learning Empirical Conditional Dist...cCorrGAN: Conditional Correlation GAN for Learning Empirical Conditional Dist...
cCorrGAN: Conditional Correlation GAN for Learning Empirical Conditional Dist...Gautier Marti
250 views27 slides

More Related Content

What's hot

유전통계학의 오늘과 내일(고려대학교 특강) by
유전통계학의 오늘과 내일(고려대학교 특강)유전통계학의 오늘과 내일(고려대학교 특강)
유전통계학의 오늘과 내일(고려대학교 특강)Woosung Yang
7K views117 slides
Exploratory data analysis using xgboost package in R by
Exploratory data analysis using xgboost package in RExploratory data analysis using xgboost package in R
Exploratory data analysis using xgboost package in RSatoshi Kato
4.3K views93 slides
Nonlinear component analysis as a kernel eigenvalue problem by
Nonlinear component analysis as a kernel eigenvalue problemNonlinear component analysis as a kernel eigenvalue problem
Nonlinear component analysis as a kernel eigenvalue problemMichele Filannino
4.2K views43 slides
CART Classification and Regression Trees Experienced User Guide by
CART Classification and Regression Trees Experienced User GuideCART Classification and Regression Trees Experienced User Guide
CART Classification and Regression Trees Experienced User GuideSalford Systems
4.8K views55 slides
Data visualization by
Data visualizationData visualization
Data visualizationMaheshwor Shrestha
835 views7 slides
Hyperparameter Optimization for Machine Learning by
Hyperparameter Optimization for Machine LearningHyperparameter Optimization for Machine Learning
Hyperparameter Optimization for Machine LearningFrancesco Casalegno
1.1K views21 slides

What's hot(20)

유전통계학의 오늘과 내일(고려대학교 특강) by Woosung Yang
유전통계학의 오늘과 내일(고려대학교 특강)유전통계학의 오늘과 내일(고려대학교 특강)
유전통계학의 오늘과 내일(고려대학교 특강)
Woosung Yang7K views
Exploratory data analysis using xgboost package in R by Satoshi Kato
Exploratory data analysis using xgboost package in RExploratory data analysis using xgboost package in R
Exploratory data analysis using xgboost package in R
Satoshi Kato4.3K views
Nonlinear component analysis as a kernel eigenvalue problem by Michele Filannino
Nonlinear component analysis as a kernel eigenvalue problemNonlinear component analysis as a kernel eigenvalue problem
Nonlinear component analysis as a kernel eigenvalue problem
Michele Filannino4.2K views
CART Classification and Regression Trees Experienced User Guide by Salford Systems
CART Classification and Regression Trees Experienced User GuideCART Classification and Regression Trees Experienced User Guide
CART Classification and Regression Trees Experienced User Guide
Salford Systems4.8K views
Hyperparameter Optimization for Machine Learning by Francesco Casalegno
Hyperparameter Optimization for Machine LearningHyperparameter Optimization for Machine Learning
Hyperparameter Optimization for Machine Learning
Francesco Casalegno1.1K views
Overfitting & Underfitting by SOUMIT KAR
Overfitting & UnderfittingOverfitting & Underfitting
Overfitting & Underfitting
SOUMIT KAR1.7K views
Tiger graph 2021 corporate overview [read only] by ercan5
Tiger graph 2021 corporate overview [read only]Tiger graph 2021 corporate overview [read only]
Tiger graph 2021 corporate overview [read only]
ercan5333 views
Beginners Guide to Non-Negative Matrix Factorization by Benjamin Bengfort
Beginners Guide to Non-Negative Matrix FactorizationBeginners Guide to Non-Negative Matrix Factorization
Beginners Guide to Non-Negative Matrix Factorization
Benjamin Bengfort33.1K views
Graph Database Meetup in Korea #4. 그래프 이론을 적용한 그래프 데이터베이스 활용 사례 by bitnineglobal
Graph Database Meetup in Korea #4. 그래프 이론을 적용한 그래프 데이터베이스 활용 사례 Graph Database Meetup in Korea #4. 그래프 이론을 적용한 그래프 데이터베이스 활용 사례
Graph Database Meetup in Korea #4. 그래프 이론을 적용한 그래프 데이터베이스 활용 사례
bitnineglobal2.7K views
Svm Presentation by shahparin
Svm PresentationSvm Presentation
Svm Presentation
shahparin1.1K views
JPL’s Institutional Knowledge Graph II: A Foundation for Constructing Enterpr... by Enterprise Knowledge
JPL’s Institutional Knowledge Graph II: A Foundation for Constructing Enterpr...JPL’s Institutional Knowledge Graph II: A Foundation for Constructing Enterpr...
JPL’s Institutional Knowledge Graph II: A Foundation for Constructing Enterpr...
Knowledge Graph for Machine Learning and Data Science by Cambridge Semantics
Knowledge Graph for Machine Learning and Data ScienceKnowledge Graph for Machine Learning and Data Science
Knowledge Graph for Machine Learning and Data Science
RM-CVaR: Regularized Multiple β-CVaR Portfolio(IJCAI Presentation) by Kei Nakagawa
RM-CVaR: Regularized Multiple β-CVaR Portfolio(IJCAI Presentation)RM-CVaR: Regularized Multiple β-CVaR Portfolio(IJCAI Presentation)
RM-CVaR: Regularized Multiple β-CVaR Portfolio(IJCAI Presentation)
Kei Nakagawa591 views
Graph Gurus 15: Introducing TigerGraph 2.4 by TigerGraph
Graph Gurus 15: Introducing TigerGraph 2.4 Graph Gurus 15: Introducing TigerGraph 2.4
Graph Gurus 15: Introducing TigerGraph 2.4
TigerGraph395 views
Graph Database Meetup in Korea #8. Graph Database 5 Offerings_ DecisionTutor ... by bitnineglobal
Graph Database Meetup in Korea #8. Graph Database 5 Offerings_ DecisionTutor ...Graph Database Meetup in Korea #8. Graph Database 5 Offerings_ DecisionTutor ...
Graph Database Meetup in Korea #8. Graph Database 5 Offerings_ DecisionTutor ...
bitnineglobal589 views

Viewers also liked

Optimal Transport between Copulas for Clustering Time Series by
Optimal Transport between Copulas for Clustering Time SeriesOptimal Transport between Copulas for Clustering Time Series
Optimal Transport between Copulas for Clustering Time SeriesGautier Marti
30.2K views25 slides
A closer look at correlations by
A closer look at correlationsA closer look at correlations
A closer look at correlationsGautier Marti
843 views51 slides
Proximal Splitting and Optimal Transport by
Proximal Splitting and Optimal TransportProximal Splitting and Optimal Transport
Proximal Splitting and Optimal TransportGabriel Peyré
1.4K views143 slides
Low Complexity Regularization of Inverse Problems - Course #3 Proximal Splitt... by
Low Complexity Regularization of Inverse Problems - Course #3 Proximal Splitt...Low Complexity Regularization of Inverse Problems - Course #3 Proximal Splitt...
Low Complexity Regularization of Inverse Problems - Course #3 Proximal Splitt...Gabriel Peyré
2.2K views94 slides
NSO_cv_20160511 by
NSO_cv_20160511NSO_cv_20160511
NSO_cv_20160511Matthew Penn
322 views10 slides
Diapo bourse aux sports by
Diapo bourse aux sportsDiapo bourse aux sports
Diapo bourse aux sportsmfrfye
275 views6 slides

Viewers also liked(19)

Optimal Transport between Copulas for Clustering Time Series by Gautier Marti
Optimal Transport between Copulas for Clustering Time SeriesOptimal Transport between Copulas for Clustering Time Series
Optimal Transport between Copulas for Clustering Time Series
Gautier Marti30.2K views
A closer look at correlations by Gautier Marti
A closer look at correlationsA closer look at correlations
A closer look at correlations
Gautier Marti843 views
Proximal Splitting and Optimal Transport by Gabriel Peyré
Proximal Splitting and Optimal TransportProximal Splitting and Optimal Transport
Proximal Splitting and Optimal Transport
Gabriel Peyré1.4K views
Low Complexity Regularization of Inverse Problems - Course #3 Proximal Splitt... by Gabriel Peyré
Low Complexity Regularization of Inverse Problems - Course #3 Proximal Splitt...Low Complexity Regularization of Inverse Problems - Course #3 Proximal Splitt...
Low Complexity Regularization of Inverse Problems - Course #3 Proximal Splitt...
Gabriel Peyré2.2K views
Diapo bourse aux sports by mfrfye
Diapo bourse aux sportsDiapo bourse aux sports
Diapo bourse aux sports
mfrfye275 views
Here be dragons by deelay1
Here be dragonsHere be dragons
Here be dragons
deelay1765 views
On the stability of clustering financial time series by Gautier Marti
On the stability of clustering financial time seriesOn the stability of clustering financial time series
On the stability of clustering financial time series
Gautier Marti861 views
Neurological considerations by Jess Sarabia
Neurological considerationsNeurological considerations
Neurological considerations
Jess Sarabia795 views
Searching for the grey gold - 2013 by Olle Bergendahl
Searching for the grey gold - 2013Searching for the grey gold - 2013
Searching for the grey gold - 2013
Olle Bergendahl341 views
Carla Casilli - Cineca + open badges - May 2015 by Bestr
Carla Casilli - Cineca + open badges - May 2015Carla Casilli - Cineca + open badges - May 2015
Carla Casilli - Cineca + open badges - May 2015
Bestr650 views
Cormac Ferrick Sociology 204 Final Presentation by Mac Ferrick
Cormac Ferrick Sociology 204 Final PresentationCormac Ferrick Sociology 204 Final Presentation
Cormac Ferrick Sociology 204 Final Presentation
Mac Ferrick205 views

Similar to Optimal Transport vs. Fisher-Rao distance between Copulas

Clustering Random Walk Time Series by
Clustering Random Walk Time SeriesClustering Random Walk Time Series
Clustering Random Walk Time SeriesGautier Marti
710 views25 slides
Bayesian phylogenetic inference_big4_ws_2016-10-10 by
Bayesian phylogenetic inference_big4_ws_2016-10-10Bayesian phylogenetic inference_big4_ws_2016-10-10
Bayesian phylogenetic inference_big4_ws_2016-10-10FredrikRonquist
1.1K views90 slides
Continuous and Discrete-Time Analysis of SGD by
Continuous and Discrete-Time Analysis of SGDContinuous and Discrete-Time Analysis of SGD
Continuous and Discrete-Time Analysis of SGDValentin De Bortoli
234 views4 slides
MUMS Opening Workshop - Panel Discussion: Facts About Some Statisitcal Models... by
MUMS Opening Workshop - Panel Discussion: Facts About Some Statisitcal Models...MUMS Opening Workshop - Panel Discussion: Facts About Some Statisitcal Models...
MUMS Opening Workshop - Panel Discussion: Facts About Some Statisitcal Models...The Statistical and Applied Mathematical Sciences Institute
95 views43 slides
Maximum likelihood estimation of regularisation parameters in inverse problem... by
Maximum likelihood estimation of regularisation parameters in inverse problem...Maximum likelihood estimation of regularisation parameters in inverse problem...
Maximum likelihood estimation of regularisation parameters in inverse problem...Valentin De Bortoli
251 views35 slides
The role of kalman filter in improving the accuracy of gps kinematic technique by
The role of kalman filter in improving the accuracy of gps kinematic techniqueThe role of kalman filter in improving the accuracy of gps kinematic technique
The role of kalman filter in improving the accuracy of gps kinematic techniqueIAEME Publication
364 views9 slides

Similar to Optimal Transport vs. Fisher-Rao distance between Copulas(20)

Clustering Random Walk Time Series by Gautier Marti
Clustering Random Walk Time SeriesClustering Random Walk Time Series
Clustering Random Walk Time Series
Gautier Marti710 views
Bayesian phylogenetic inference_big4_ws_2016-10-10 by FredrikRonquist
Bayesian phylogenetic inference_big4_ws_2016-10-10Bayesian phylogenetic inference_big4_ws_2016-10-10
Bayesian phylogenetic inference_big4_ws_2016-10-10
FredrikRonquist1.1K views
Maximum likelihood estimation of regularisation parameters in inverse problem... by Valentin De Bortoli
Maximum likelihood estimation of regularisation parameters in inverse problem...Maximum likelihood estimation of regularisation parameters in inverse problem...
Maximum likelihood estimation of regularisation parameters in inverse problem...
The role of kalman filter in improving the accuracy of gps kinematic technique by IAEME Publication
The role of kalman filter in improving the accuracy of gps kinematic techniqueThe role of kalman filter in improving the accuracy of gps kinematic technique
The role of kalman filter in improving the accuracy of gps kinematic technique
IAEME Publication364 views
random forests for ABC model choice and parameter estimation by Christian Robert
random forests for ABC model choice and parameter estimationrandom forests for ABC model choice and parameter estimation
random forests for ABC model choice and parameter estimation
Christian Robert3.6K views
Natalini nse slide_giu2013 by Madd Maths
Natalini nse slide_giu2013Natalini nse slide_giu2013
Natalini nse slide_giu2013
Madd Maths648 views
Q-Metrics in Theory and Practice by Magdi Mohamed
Q-Metrics in Theory and PracticeQ-Metrics in Theory and Practice
Q-Metrics in Theory and Practice
Magdi Mohamed304 views
Q-Metrics in Theory And Practice by guest3550292
Q-Metrics in Theory And PracticeQ-Metrics in Theory And Practice
Q-Metrics in Theory And Practice
guest3550292248 views

More from Gautier Marti

Using Large Language Models in 10 Lines of Code by
Using Large Language Models in 10 Lines of CodeUsing Large Language Models in 10 Lines of Code
Using Large Language Models in 10 Lines of CodeGautier Marti
1.3K views13 slides
What deep learning can bring to... by
What deep learning can bring to...What deep learning can bring to...
What deep learning can bring to...Gautier Marti
94 views12 slides
A quick demo of Top2Vec With application on 2020 10-K business descriptions by
A quick demo of Top2Vec With application on 2020 10-K business descriptionsA quick demo of Top2Vec With application on 2020 10-K business descriptions
A quick demo of Top2Vec With application on 2020 10-K business descriptionsGautier Marti
523 views15 slides
How deep generative models can help quants reduce the risk of overfitting? by
How deep generative models can help quants reduce the risk of overfitting?How deep generative models can help quants reduce the risk of overfitting?
How deep generative models can help quants reduce the risk of overfitting?Gautier Marti
3.4K views59 slides
Generating Realistic Synthetic Data in Finance by
Generating Realistic Synthetic Data in FinanceGenerating Realistic Synthetic Data in Finance
Generating Realistic Synthetic Data in FinanceGautier Marti
3.9K views64 slides
Applications of GANs in Finance by
Applications of GANs in FinanceApplications of GANs in Finance
Applications of GANs in FinanceGautier Marti
3K views61 slides

More from Gautier Marti(16)

Using Large Language Models in 10 Lines of Code by Gautier Marti
Using Large Language Models in 10 Lines of CodeUsing Large Language Models in 10 Lines of Code
Using Large Language Models in 10 Lines of Code
Gautier Marti1.3K views
What deep learning can bring to... by Gautier Marti
What deep learning can bring to...What deep learning can bring to...
What deep learning can bring to...
Gautier Marti94 views
A quick demo of Top2Vec With application on 2020 10-K business descriptions by Gautier Marti
A quick demo of Top2Vec With application on 2020 10-K business descriptionsA quick demo of Top2Vec With application on 2020 10-K business descriptions
A quick demo of Top2Vec With application on 2020 10-K business descriptions
Gautier Marti523 views
How deep generative models can help quants reduce the risk of overfitting? by Gautier Marti
How deep generative models can help quants reduce the risk of overfitting?How deep generative models can help quants reduce the risk of overfitting?
How deep generative models can help quants reduce the risk of overfitting?
Gautier Marti3.4K views
Generating Realistic Synthetic Data in Finance by Gautier Marti
Generating Realistic Synthetic Data in FinanceGenerating Realistic Synthetic Data in Finance
Generating Realistic Synthetic Data in Finance
Gautier Marti3.9K views
Applications of GANs in Finance by Gautier Marti
Applications of GANs in FinanceApplications of GANs in Finance
Applications of GANs in Finance
Gautier Marti3K views
My recent attempts at using GANs for simulating realistic stocks returns by Gautier Marti
My recent attempts at using GANs for simulating realistic stocks returnsMy recent attempts at using GANs for simulating realistic stocks returns
My recent attempts at using GANs for simulating realistic stocks returns
Gautier Marti1.1K views
Takeaways from ICML 2019, Long Beach, California by Gautier Marti
Takeaways from ICML 2019, Long Beach, CaliforniaTakeaways from ICML 2019, Long Beach, California
Takeaways from ICML 2019, Long Beach, California
Gautier Marti1.3K views
A review of two decades of correlations, hierarchies, networks and clustering... by Gautier Marti
A review of two decades of correlations, hierarchies, networks and clustering...A review of two decades of correlations, hierarchies, networks and clustering...
A review of two decades of correlations, hierarchies, networks and clustering...
Gautier Marti1.2K views
Autoregressive Convolutional Neural Networks for Asynchronous Time Series by Gautier Marti
Autoregressive Convolutional Neural Networks for Asynchronous Time SeriesAutoregressive Convolutional Neural Networks for Asynchronous Time Series
Autoregressive Convolutional Neural Networks for Asynchronous Time Series
Gautier Marti968 views
Some contributions to the clustering of financial time series - Applications ... by Gautier Marti
Some contributions to the clustering of financial time series - Applications ...Some contributions to the clustering of financial time series - Applications ...
Some contributions to the clustering of financial time series - Applications ...
Gautier Marti1.6K views
Clustering CDS: algorithms, distances, stability and convergence rates by Gautier Marti
Clustering CDS: algorithms, distances, stability and convergence ratesClustering CDS: algorithms, distances, stability and convergence rates
Clustering CDS: algorithms, distances, stability and convergence rates
Gautier Marti662 views
Clustering Financial Time Series using their Correlations and their Distribut... by Gautier Marti
Clustering Financial Time Series using their Correlations and their Distribut...Clustering Financial Time Series using their Correlations and their Distribut...
Clustering Financial Time Series using their Correlations and their Distribut...
Gautier Marti523 views
Clustering Financial Time Series: How Long is Enough? by Gautier Marti
Clustering Financial Time Series: How Long is Enough?Clustering Financial Time Series: How Long is Enough?
Clustering Financial Time Series: How Long is Enough?
Gautier Marti700 views
On Clustering Financial Time Series - Beyond Correlation by Gautier Marti
On Clustering Financial Time Series - Beyond CorrelationOn Clustering Financial Time Series - Beyond Correlation
On Clustering Financial Time Series - Beyond Correlation
Gautier Marti614 views
On clustering financial time series - A need for distances between dependent ... by Gautier Marti
On clustering financial time series - A need for distances between dependent ...On clustering financial time series - A need for distances between dependent ...
On clustering financial time series - A need for distances between dependent ...
Gautier Marti990 views

Recently uploaded

Data about the sector workshop by
Data about the sector workshopData about the sector workshop
Data about the sector workshopinfo828217
15 views27 slides
[DSC Europe 23] Spela Poklukar & Tea Brasanac - Retrieval Augmented Generation by
[DSC Europe 23] Spela Poklukar & Tea Brasanac - Retrieval Augmented Generation[DSC Europe 23] Spela Poklukar & Tea Brasanac - Retrieval Augmented Generation
[DSC Europe 23] Spela Poklukar & Tea Brasanac - Retrieval Augmented GenerationDataScienceConferenc1
15 views29 slides
apple.pptx by
apple.pptxapple.pptx
apple.pptxhoneybeeqwe
5 views15 slides
[DSC Europe 23][AI:CSI] Aleksa Stojanovic - Applying AI for Threat Detection ... by
[DSC Europe 23][AI:CSI] Aleksa Stojanovic - Applying AI for Threat Detection ...[DSC Europe 23][AI:CSI] Aleksa Stojanovic - Applying AI for Threat Detection ...
[DSC Europe 23][AI:CSI] Aleksa Stojanovic - Applying AI for Threat Detection ...DataScienceConferenc1
6 views15 slides
CRM stick or twist.pptx by
CRM stick or twist.pptxCRM stick or twist.pptx
CRM stick or twist.pptxinfo828217
11 views16 slides
CRIJ4385_Death Penalty_F23.pptx by
CRIJ4385_Death Penalty_F23.pptxCRIJ4385_Death Penalty_F23.pptx
CRIJ4385_Death Penalty_F23.pptxyvettemm100
7 views24 slides

Recently uploaded(20)

Data about the sector workshop by info828217
Data about the sector workshopData about the sector workshop
Data about the sector workshop
info82821715 views
[DSC Europe 23] Spela Poklukar & Tea Brasanac - Retrieval Augmented Generation by DataScienceConferenc1
[DSC Europe 23] Spela Poklukar & Tea Brasanac - Retrieval Augmented Generation[DSC Europe 23] Spela Poklukar & Tea Brasanac - Retrieval Augmented Generation
[DSC Europe 23] Spela Poklukar & Tea Brasanac - Retrieval Augmented Generation
[DSC Europe 23][AI:CSI] Aleksa Stojanovic - Applying AI for Threat Detection ... by DataScienceConferenc1
[DSC Europe 23][AI:CSI] Aleksa Stojanovic - Applying AI for Threat Detection ...[DSC Europe 23][AI:CSI] Aleksa Stojanovic - Applying AI for Threat Detection ...
[DSC Europe 23][AI:CSI] Aleksa Stojanovic - Applying AI for Threat Detection ...
CRM stick or twist.pptx by info828217
CRM stick or twist.pptxCRM stick or twist.pptx
CRM stick or twist.pptx
info82821711 views
CRIJ4385_Death Penalty_F23.pptx by yvettemm100
CRIJ4385_Death Penalty_F23.pptxCRIJ4385_Death Penalty_F23.pptx
CRIJ4385_Death Penalty_F23.pptx
yvettemm1007 views
OECD-Persol Holdings Workshop on Advancing Employee Well-being in Business an... by StatsCommunications
OECD-Persol Holdings Workshop on Advancing Employee Well-being in Business an...OECD-Persol Holdings Workshop on Advancing Employee Well-being in Business an...
OECD-Persol Holdings Workshop on Advancing Employee Well-being in Business an...
Survey on Factuality in LLM's.pptx by NeethaSherra1
Survey on Factuality in LLM's.pptxSurvey on Factuality in LLM's.pptx
Survey on Factuality in LLM's.pptx
NeethaSherra17 views
Data Journeys Hard Talk workshop final.pptx by info828217
Data Journeys Hard Talk workshop final.pptxData Journeys Hard Talk workshop final.pptx
Data Journeys Hard Talk workshop final.pptx
info82821710 views
Short Story Assignment by Kelly Nguyen by kellynguyen01
Short Story Assignment by Kelly NguyenShort Story Assignment by Kelly Nguyen
Short Story Assignment by Kelly Nguyen
kellynguyen0119 views
[DSC Europe 23][Cryptica] Martin_Summer_Digital_central_bank_money_Ideas_init... by DataScienceConferenc1
[DSC Europe 23][Cryptica] Martin_Summer_Digital_central_bank_money_Ideas_init...[DSC Europe 23][Cryptica] Martin_Summer_Digital_central_bank_money_Ideas_init...
[DSC Europe 23][Cryptica] Martin_Summer_Digital_central_bank_money_Ideas_init...
Cross-network in Google Analytics 4.pdf by GA4 Tutorials
Cross-network in Google Analytics 4.pdfCross-network in Google Analytics 4.pdf
Cross-network in Google Analytics 4.pdf
GA4 Tutorials6 views
[DSC Europe 23] Danijela Horak - The Innovator’s Dilemma: to Build or Not to ... by DataScienceConferenc1
[DSC Europe 23] Danijela Horak - The Innovator’s Dilemma: to Build or Not to ...[DSC Europe 23] Danijela Horak - The Innovator’s Dilemma: to Build or Not to ...
[DSC Europe 23] Danijela Horak - The Innovator’s Dilemma: to Build or Not to ...
PRIVACY AWRE PERSONAL DATA STORAGE by antony420421
PRIVACY AWRE PERSONAL DATA STORAGEPRIVACY AWRE PERSONAL DATA STORAGE
PRIVACY AWRE PERSONAL DATA STORAGE
antony4204215 views
Chapter 3b- Process Communication (1) (1)(1) (1).pptx by ayeshabaig2004
Chapter 3b- Process Communication (1) (1)(1) (1).pptxChapter 3b- Process Communication (1) (1)(1) (1).pptx
Chapter 3b- Process Communication (1) (1)(1) (1).pptx
ayeshabaig20047 views
[DSC Europe 23] Predrag Ilic & Simeon Rilling - From Data Lakes to Data Mesh ... by DataScienceConferenc1
[DSC Europe 23] Predrag Ilic & Simeon Rilling - From Data Lakes to Data Mesh ...[DSC Europe 23] Predrag Ilic & Simeon Rilling - From Data Lakes to Data Mesh ...
[DSC Europe 23] Predrag Ilic & Simeon Rilling - From Data Lakes to Data Mesh ...

Optimal Transport vs. Fisher-Rao distance between Copulas

  • 1. Introduction Statistical distances Optimal Transport vs. Fisher-Rao distance between Copulas IEEE SSP 2016 G. Marti, S. Andler, F. Nielsen, P. Donnat June 28, 2016 Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
  • 2. Introduction Statistical distances Clustering of Time Series We need a distance Dij between time series xi and xj If we look for ‘correlation’, Dij is a decreasing function of ρij , a measure of ‘correlation’ Several choices are available for ρij . . . Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
  • 3. Introduction Statistical distances Copulas Sklar’s Theorem: F(xi , xj ) = Cij (Fi (xi ), Fj (xj )) Cij , the copula, encodes the dependence structure Fr´echet-Hoeffding bounds: max{ui + uj − 1, 0} ≤ Cij (ui , uj ) ≤ min{ui , uj } (left) lower-bound, (mid) independence, (right) upper-bound copulas Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
  • 4. Introduction Statistical distances Copulas - Gaussian Example Gaussian copula: CGauss R (ui , uj ) = ΦR(Φ−1(ui ), Φ−1(uj )) The distribution is parametrized by a correlation matrix R. Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
  • 5. Introduction Statistical distances The Target/Forget (copula-based) Dependence Coefficient Dependence is measured as the relative distance from independence to the nearest target-dependence: comonotonicity or counter-monotonicity Which distances are appropriate between copulas for the task of clustering (copulas and time series)? Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
  • 6. Introduction Statistical distances Definitions - Fisher-Rao geodesic distance Metrization of the paramater space {θ ∈ Rd | p(X; θ)dx = 1}. Consider the metric gjk(θ) = − ∂2 log p(x,θ) ∂θj ∂θk p(x, θ)dx, the infinitesimal length ds(θ) = ( θ) G(θ) θ, the Fisher-Rao geodesic distance FR(θ1, θ2) = θ2 θ1 ds(θ). f -divergences induce infinitesimal length proportional to Fisher-Rao infinitesimal length: Df (θ θ + dθ) = 1 2 ( θ) G(θ) θ. Thus, they have the same local behaviour [1]. Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
  • 7. Introduction Statistical distances Definitions - Optimal Transport distances Wasserstein metric Wp(µ, ν)p = inf γ∈Γ(µ,ν) M×M d(x, y)p dγ(x, y) Image from Optimal Transport for Image Processing, Papadakis Other transportation distances: regularized discrete optimal transport [3], Sinkhorn distances [2], . . . Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
  • 8. Introduction Statistical distances Geometry of covariances Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
  • 9. Introduction Statistical distances Distances between Gaussian copulas Copulas C1, C2, C3 encoding a correlation of 0.5, 0.99, 0.9999 respectively; Which pair of copulas is the nearest? - For Fisher-Rao, Kullback-Leibler, Hellinger and related divergences: D(C1, C2) ≤ D(C2, C3); - For Wasserstein: W2(C2, C3) ≤ W2(C1, C2) Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
  • 10. Introduction Statistical distances Distances as a function of (ρ1, ρ2) Distance heatmap and surface as a function of (ρ1, ρ2) for Fisher-Rao for Wasserstein W2 Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
  • 11. Introduction Statistical distances Distances impact on clustering Datasets of bivariate time series are generated from six Gaussian copulas with correlation .1, .2, .6, .7, .99, .9999 Distance heatmaps for Fisher-Rao (left), W2 (right); Using Ward clustering, Fisher-Rao yields clusters of copulas with correlations {.1, .2, .6, .7}, {.99}, {.9999}, W2 yields {.1, .2}, {.6, .7}, {.99, .9999} Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
  • 12. Introduction Statistical distances Fisher metric and the Cram´er–Rao lower bound Cram´er–Rao lower bound (CRLB) The variance of any unbiased estimator ˆθ of θ is bounded by the reciprocal of the Fisher information G(θ): var(ˆθ) ≥ 1 G(θ) . In the bivariate Gaussian copula case, var(ˆρ) ≥ (ρ − 1)2(ρ + 1)2 3(ρ2 + 1) . Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
  • 13. Introduction Statistical distances Fisher metric and the Cram´er–Rao lower bound We consider the set of 2 × 2 correlation matrices C = 1 θ θ 1 parameterized by θ. Let x = x1 x2 ∈ R2 . f (x; θ) = 1 2π 1−θ2 exp − 1 2 x C−1 x = 1 2π 1−θ2 exp − 1 2(1−θ2) (x2 1 + x2 2 − 2θx1x2) log f (x; θ) = − log(2π 1 − θ2) − 1 2(1−θ2) (x2 1 + x2 2 − 2θx1x2) ∂2 log f (x;θ) ∂θ2 = − θ2+1 (θ2−1)2 − x2 1 2(θ+1)3 + x2 1 2(θ−1)3 − x2 2 2(θ+1)3 + x2 2 2(θ−1)3 − x1x2 (θ+1)3 − x1x2 (θ−1)3 Then, we compute ∞ −∞ ∂2 log f (x;θ) ∂θ2 f (x; θ)dx. Since E[x1] = E[x2] = 0, E[x1x2] = θ, E[x2 1 ] = E[x2 2 ] = 1, we get ∞ −∞ ∂2 log f (x;θ) ∂θ2 f (x; θ)dx = − θ2+1 (θ2−1)2 − 1 2(θ+1)3 + 1 2(θ−1)3 − 1 2(θ+1)3 + 1 2(θ−1)3 − θ (θ+1)3 − θ (θ−1)3 = − 3(θ2+1) (θ−1)2(θ+1)2 Thus, G(θ) = 3(θ2 + 1) (θ − 1)2(θ + 1)2 . Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
  • 14. Introduction Statistical distances Fisher metric and the Cram´er–Rao lower bound In the bivariate Gaussian copula case, var(ˆρ) ≥ (ρ − 1)2(ρ + 1)2 3(ρ2 + 1) . Recall that locally Fisher-Rao and the f -divergences are a quadratic form of the Fisher metric ( θ) G(θ) θ. So, the discriminative power of these distances is well calibrated with respect to statistical uncertainty. For this purpose, they induce the appropriate curvature on the parameter space. Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
  • 15. Introduction Statistical distances Properties of these distances In addition, for clustering we prefer OT since: in a parametric setting: Fisher-Rao and f -divergences are defined on density manifolds, but some important copulas (such as the Fr´echet-Hoeffding upper bound) do not belong to these manifolds; Thus, in case of closed-form formulas (such as in the Gaussian case), they are ill-defined for these copulas (for perfect dependence, covariance is not invertible) in a non-parametric/empirical setting: f -divergences are defined for absolutely continuous measures, thus require a pre-processing KDE they are not aware of the support geometry, thus badly handle noise on the support Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
  • 16. Introduction Statistical distances Barycenters OT is defined for both discrete/empirical and continuous measures and is support-geometry aware: 0 0.5 1 0 0.5 1 0.0000 0.0015 0.0030 0.0045 0.0060 0.0075 0.0090 0.0105 0.0120 0 0.5 1 0 0.5 1 0.0000 0.0015 0.0030 0.0045 0.0060 0.0075 0.0090 0.0105 0.0120 0 0.5 1 0 0.5 1 0.0000 0.0008 0.0016 0.0024 0.0032 0.0040 0.0048 0.0056 0 0.5 1 0 0.5 1 0.0000 0.0015 0.0030 0.0045 0.0060 0.0075 0.0090 0.0105 0.0120 0 0.5 1 0 0.5 1 0.0000 0.0015 0.0030 0.0045 0.0060 0.0075 0.0090 0.0105 0.0120 5 copulas describing the dependence between X ∼ U([0, 1]) and Y ∼ (X ± i )2 , where i is a constant noise specific for each distribution 0 0.5 1 0 0.5 1 Wasserstein barycenter copula 0.0000 0.0004 0.0008 0.0012 0.0016 0.0020 0.0024 0.0028 0.0032 Barycenter of the 5 copulas for a divergence and OT Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
  • 17. Introduction Statistical distances Future Research Develop further geometries of copulas using Optimal Transport: show that dependence-clustering of time series is improved over standard correlations using f -divergences: detect efficiently dependence-regime switching in multivariate time series (cf. Fr´ed´eric Barbaresco’s work on radar signal processing) Numerical experiments and code: https://www.datagrapple.com/Tech/fisher-vs-ot.html Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas
  • 18. Introduction Statistical distances Shun-ichi Amari and Andrzej Cichocki. Information geometry of divergence functions. Bulletin of the Polish Academy of Sciences: Technical Sciences, 58(1):183–195, 2010. Marco Cuturi. Sinkhorn distances: Lightspeed computation of optimal transport. In Advances in Neural Information Processing Systems, pages 2292–2300, 2013. Sira Ferradans, Nicolas Papadakis, Julien Rabin, Gabriel Peyr´e, and Jean-Fran¸cois Aujol. Regularized discrete optimal transport. Springer, 2013. Gautier Marti Optimal Transport vs. Fisher-Rao distance between Copulas