Subscribe to DSC Newsletter
Antoine Savine
  • Male
  • Copenhagen
  • Denmark
Share on Facebook
Share

RSS

Differential Machine Learning: 5min video overview

Lightning talk - Bloomberg BBQ 28th May 2020 - main ideas of differential ML with application to derivatives pricing and risk management

QuantMinds e-magazine, June 2020

Differential machine learning combines ML with automatic differentiation (AAD) to produce accurate pricing and risk approximations for arbitrary derivatives transactions or trading books, quickly, online, with convergence guarantees.

Deep Pricing Webinar

Differential ML is applicable in situations where high quality first order derivatives wrt training inputs are available. In the context of financial Derivatives and risk management, pathwise differentials are efficiently computed with automatic adjoint differentiation (AAD). Differential machine learning, combined with AAD, provides extremely effective pricing and risk approximations. We can produce fast pricing analytics in models too complex for closed form solutions, extract the risk factors of complex transactions and trading books, and effectively compute risk management metrics like reports across a large number of scenarios, backtesting and simulation of hedge strategies, or regulations like XVA, CCR, FRTB or SIMM-MVA.

Differential Machine Learning

Differential machine learning (ML) is an extension of supervised learning, where ML models are trained on examples of not only inputs and labels but also differentials of labels to inputs. Differential ML is applicable in situations where high quality first order derivatives wrt training inputs are available. In the context of financial Derivatives and risk management, pathwise differentials are efficiently computed with automatic adjoint differentiation (AAD). Differential machine learning, combined with AAD, provides extremely effective pricing and risk approximations. We can produce fast pricing analytics in models too complex for closed form solutions, extract the risk factors of complex transactions and trading books, and effectively compute risk management metrics like reports across a large number of scenarios, backtesting and simulation of hedge strategies, or regulations like XVA, CCR, FRTB or SIMM-MVA. The article focuses on differential deep learning (DL), arguably the strongest application. Standard DL trains neural networks (NN) on punctual examples, whereas differential DL teaches them the shape of the target function, resulting in vastly improved performance, illustrated with a number of numerical examples, both idealized and real world. In the online appendices, we apply differential learning to other ML models, like classic regression or principal component analysis (PCA), with equally remarkable results. We also posted a TensorFlow implementation notebook designed to run on Google Colab.

Article with code, Wilmott Mar20

A. Savine, “Computation graphs for aad and machine learning part iii: application to derivatives sensitivities” Wilmott, vol. 2020, iss. 106, p. 24-39, 2020.

Gifts Received

Gift

Antoine Savine has not received any gifts yet

Give a Gift

 

Antoine Savine's Page

Latest Activity

Antoine Savine posted a video

Differential Machine Learning 5min Video Overview -- Antoine Savine

In this lightning talk delivered fro Bloomberg's BBQ seminar 28th May 2020, we expose the main ideas of differential machine learning and application to deri...
Jun 14
Antoine Savine posted a video

AAD & Backpropagation in Machine Learning and Finance, Explained in 15min -- Antoine Savine

In this lightning talk, we introduce the main ideas of adjoint differentiation and backpropagation technologies, with application in machine learning and finance
Jun 4
Antoine Savine posted a blog post

Differential ML on TensorFlow and Colab

Brian Huge and I just posted a working paper following six months of research and development on function approximation by artificial intelligence (AI) in Danske Bank. One major finding was that training machine learning (ML) models for regression (i.e. prediction…See More
May 26
Antoine Savine updated their profile
May 25
Pendekanti Mahesh Babu liked Antoine Savine's blog post Differential ML on TensorFlow and Colab
May 12
Pendekanti Mahesh Babu liked Antoine Savine's blog post Differential ML on TensorFlow and Colab
May 12
Antoine Savine's blog post was featured

Differential ML on TensorFlow and Colab

Brian Huge and I just posted a working paper following six months of research and development on function approximation by artificial intelligence (AI) in Danske Bank. One major finding was that training machine learning (ML) models for regression (i.e. prediction…See More
May 10
Antoine Savine posted a blog post
Apr 19
farooq hassan liked Antoine Savine's blog post Deep Analytics: Risk Management with AI
Dec 12, 2019
Antoine Savine's blog post was featured

Deep Analytics: Risk Management with AI

We first provide a mini-tutorial on  Adjoint Algorithmic Differentiation (AAD) (also known as back-propagation in machine learning). We then illustrate how  neural networks may be used to compute dynamic values and risks of trading books with applications to risk management of derivatives,  valuation adjustments (XVA), counterpart credit risk, FRTB and SIMM margin valuation adjustments (MVA). We also describe new techniques to substantially improve deep learning on simulated data, and discuss…See More
Dec 12, 2019
PHAM-HI liked Antoine Savine's blog post Recorded workshop from Kings College London: AAD, Backpropagation and Machine Learning in Finance
Nov 3, 2019
Antoine Savine's blog post was featured

Recorded workshop from Kings College London: AAD, Backpropagation and Machine Learning in Finance

Automatic Adjoint Differentiation (AAD) and back-propagation are key technologies in modern machine learning and finance. It is back-prop that enables deep neural networks to learn to identify faces on photographs in reasonable time. It is AAD that allows financial institutions to compute the risks of complex derivatives books in real time. The two technologies share common roots.See the AAD book here:…See More
Oct 31, 2019

Profile Information

Company:
Danske Bank
Job Title:
Quantitative Research -- Superfly Analytics
Seniority:
Staff
Job Function:
AI/ML Models
Number of employees:
10.000 to 24.999
Industry:
Financial/Banking
Short Bio:
Antoine Savine is a leading practitioner and a lecturer in methematical and computational finance. He is the author of the Modern Computational Finance book with Wiley.
Antoine holds a MsC (mathematics) from the University of Paris-Diderot and a PhD (mathematics) from Copenhagen University. He is well known in the quantitative finance community for influential work on cash-flow scripting, multi-factor interest rate models, generalized derivatives in the context of local and stochastic volatility models, and the wide adoption of AAD in financial systems. His latest research focuses on 'Differential Machine Learning' and application in finance.
LinkedIn Profile:
http://antoinesavine.com
Interests:
Contributing, Networking

Antoine Savine's Blog

Differential ML on TensorFlow and Colab

Posted on May 25, 2020 at 11:30am 0 Comments

Brian Huge and I just posted a working paper following six months of…

Continue

Automatic Differentiation in 15 Minutes -- video tutorial with application in machine learning and finance

Posted on April 17, 2020 at 7:53am 0 Comments

Recorded in Bloomberg's London offices in November 2019:

slides here

Deep Analytics: Risk Management with AI

Posted on December 10, 2019 at 1:30am 0 Comments

We first provide a mini-tutorial on  Adjoint Algorithmic Differentiation (AAD) (also known as back-propagation in machine learning). We then illustrate how  neural networks may be used to compute dynamic values and risks of trading books with applications to risk management of derivatives,  valuation adjustments (XVA), counterpart credit risk, FRTB and SIMM margin valuation adjustments (MVA). We also describe new techniques to substantially improve deep learning on simulated data, and…

Continue

Recorded workshop from Kings College London: AAD, Backpropagation and Machine Learning in Finance

Posted on October 30, 2019 at 7:00am 0 Comments

Automatic Adjoint Differentiation (AAD) and back-propagation are key technologies in modern machine learning and finance. It is back-prop that enables deep neural networks to learn to identify faces on photographs in reasonable time. It is AAD that allows financial institutions to compute the risks of complex derivatives books in real time. The two technologies share common roots.

See the AAD book here:…

Continue

Comment Wall

You need to be a member of Data Science Central to add comments!

Join Data Science Central

  • No comments yet!

Antoine Savine's Videos

  • Add Videos
  • View All
 
 
 

Videos

  • Add Videos
  • View All

© 2020   Data Science Central ®   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service