Machine learning model validation
View AgendaKey reasons to attend
-
Gain skills to build interpretable models and control model complexity
-
Learn about local and global explainability methods
-
Understand model weakness through error slicing
Customised solutions
Does your team require a tailored learning solution on this or any other topic?
Working with the portfolio of expert tutors and Risk.net’s editorial team, we can develop and deliver a customised learning to make the most impact for your team, from initial assessment to final review.
About the course
This virtual learning course focuses on the latest developments in model validation for machine learning with special emphasis on evaluation of conceptual soundness and outcome analysis.
Led by expert speakers, participants will receive hands-on learning experiences using the free, online tool PiML.
Participants will explore how to manage problems associated with over-parameterised and under-specified problems that commonly occur in machine learning models. Sessions will focus on model weakness, reliability and robustness, and will highlight how to use PiML to solidify concepts relating to explainability and causality.
Case studies and interaction with experts will develop participants’ knowledge about machine learning model validation.
Flexible pricing options:
-
Early-bird rate: book in advance and save $200
-
3-for-2 group rate: book three delegates for the price of two and save more than $2,000
-
Season tickets: book a team of 10 or more and save up to 50%
Learning objectives
-
Evaluate conceptual soundness of machine learning
-
Design interpretable machine learning models
-
Perform model-testing to evaluate model weakness, reliability, robustness and resiliency
-
Manage model fairness and model debiasing
-
Use outcome analysis evaluation
Who should attend
Relevant departments may include but are not limited to:
-
Risk model validation
-
Model risk
-
Risk management
-
Market/credit risk management
-
Stress-testing
-
Model review
Agenda
June 21–23, 2023
Timezones: Emea/Americas
Sessions:
-
Machine learning risk
-
Introduction to machine learning and machine learning model validation
-
Post-hoc explainability
-
Feature selection and causality
-
Deep rectified linear activation unit (ReLU) networks as locally interpretable models
-
Functional ANOVA and globally interpretable models
-
Model weakness and robustness
-
Model resilience
-
Model fairness
Tutors
- Agus Sudjianto, executive vide-president and head of corporate model risk, Wells Fargo
Accreditation
This course is CPD (Continued Professional Development) accredited. One credit is awarded for every hour of learning at the event.
Pre-reading materials
The Risk.net resources below have been selected to enhance your learning experience:
-
Machine learning and AI in model risk management: a quant perspective - Read article
-
Machine learning models: the validation challenge - Read article
-
Deep hedging: learning to remove the drift - Read article
To access some of the above articles you need to have a current subscription to Risk.net. If you don’t have one now, please subscribe to a free trial
Expand your learning
Events
The premier meeting place for the risk community. Providing clarity and guidance on the fast-changing regulatory landscape of capital, credit and market risk, liquidity and derivates use.
Journals
Risk Journals deliver academically rigorous, practitioner-focused content and resources for the rapidly evolving discipline of financial risk management.
Books
Risk Books are authored by leading professionals and academics. With over 150 books spanning 1,000s of chapters, our publications team is committed to connecting readers with these world class experts.
Glossary
Regularly updated by our team analysts, journalists and producers, our glossary demystifies the terminology and acronyms used in risk management, risk transfer, policy, technology and innovation.