NGO-GM: Natural Gradient Optimization for Graphical Models

18 Pages Posted: 3 Jun 2019

See all articles by Eric Benhamou

Eric Benhamou

Université Paris Dauphine; AI For Alpha; EB AI Advisory; Université Paris-Dauphine, PSL Research University

Jamal Atif

Université Paris Dauphine

Rida Laraki

Université Paris-Dauphine, PSL Research University

David Saltiel

Université Paris Dauphine; A.I. Square Connect; AI For Alpha

Date Written: May 14, 2019

Abstract

This paper deals with estimating model parameters in graphical models. We reformulate it as an information geometric optimization problem and introduce a natural gradient descent strategy that incorporates additional meta parameters. We show that our approach is a strong alternative to the celebrated EM approach for learning in graphical models. Actually, our natural gradient based strategy leads to learning optimal parameters for the final objective function without artificially trying to fit a distribution that may not correspond to the real one. We support our theoretical findings with the question of trend detection in financial markets and show that the learned model performs better than traditional practitioner methods and is less prone to overfitting.

Keywords: Graphical Model, Natural Gradient, Overfitting

JEL Classification: C12, G11

Suggested Citation

Benhamou, Eric and Atif, Jamal and Laraki, Rida and Saltiel, David, NGO-GM: Natural Gradient Optimization for Graphical Models (May 14, 2019). Université Paris-Dauphine Research Paper No. 3387874, Available at SSRN: https://ssrn.com/abstract=3387874 or http://dx.doi.org/10.2139/ssrn.3387874

Eric Benhamou (Contact Author)

Université Paris Dauphine ( email )

Place du Maréchal de Tassigny
Paris, Cedex 16 75775
France

AI For Alpha ( email )

35 boulevard d'Inkermann
Neuilly sur Seine, 92200
France

EB AI Advisory ( email )

35 Boulevard d'Inkermann
Neuilly sur Seine, 92200
France

Université Paris-Dauphine, PSL Research University ( email )

Place du Maréchal de Lattre de Tassigny
Paris, 75016
France

Jamal Atif

Université Paris Dauphine ( email )

Place du Maréchal de Tassigny
Paris, Cedex 16 75775
France

Rida Laraki

Université Paris-Dauphine, PSL Research University ( email )

Place du Maréchal de Lattre de Tassigny
Paris, 75016
France

David Saltiel

Université Paris Dauphine ( email )

Place du Maréchal de Tassigny
Paris, Cedex 16 75775
France

A.I. Square Connect ( email )

35 Boulevard d'Inkermann
Neuilly sur Seine, 92200
France

AI For Alpha ( email )

35 boulevard d'Inkermann
Neuilly sur Seine, 92200
France

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
4,361
Abstract Views
34,569
Rank
4,838
PlumX Metrics