IA|BE is organizing for the first time in its history a training program that will offer the possibility to obtain a certificate: the IA|BE Actuarial Data Scientist Certificate.
The objective of the training is to provide to the participants a deep understanding of key Data Science methods, their domain of applicability, key results of interest for practitioners and connections with the classical actuarial methods.
Moreover, the program emphasizes on applied programming practice and demystifies concepts such as model deployment in a corporate environment, MLOps, Object-Oriented programming, code management tools (e.g. Github), and principles of software engineering. For this purpose, Python has been chosen as programming language.
The program consists of 3 modules, each of which is followed by an assignment.
Each module consists of 4 theory sessions and 4 programming sessions.
The theory sessions will be given by the professors of the KU Leuven, UC Louvain and Université Libre de Bruxelles.
For the programming sessions, the Board of IA|BE has decided to cooperate with Reacfin and a team of KU Leuven affiliated researchers in (insurance) data science, coordinated by prof. K. Antonio.
Participants who successfully complete the 3 modules will receive the Certificate of IA|BE Actuarial Data Scientist.
Each of these modules can also be taken separately.
Important remark: The detailed content of the various sessions of the modules may still be adjusted in the course of the next few days on the basis of the information we will receive from the various speakers.
Module 3: Advanced topics in Machine learning
This module extends on the learnings from the two previous modules to introduce transparency and ethics in machine learning models.
The importance and choice of variables, their (additive) contributions to the model's behavior and causal relations will be explored.
The last session will invite a panel of speakers to present new topics of interest in the machine learning community, of relevance for the actuarial community.
This session concludes by opening perspective on the problems modern machine learning can tackle, beyond classification and regression problems.
- When registering for the full programme of 3 Modules : 800 € / Module
- Otherwise : 1 000 € / Module
The sessions will be set up online at the start. When, in the course of the programme, it becomes possible to have the sessions physically (at the Actuarial House) and online at the same time, participants will be informed and will be able to choose the format in which they wish to participate.
In order to guarantee optimal training conditions and interactivity of the programming sessions, the number of participants is currently limited to 20.
Schedule of Module 3 : Advanced topics in Machine Learning
|Day 1, Tuesday 19 April|
|16:00 - 18:00||Global & Local interpretation of Machine Learning models. By HAINAUT Donatien|
|Day 2, Tuesday 26 April|
|16:00 - 18:00||Programming : Interpretability tools|
|Day 3, Tuesday 3 May|
|16:00 - 18:00||Variable selection and model agnostic methods By VAN OIRBEEK Robin|
|Day 4, Tuesday 10 May|
|16:00 - 18:00||Programming : Putting models in production (Part I)|
|Day 5, Tuesday 17 May|
|16:00 - 18:00||Causality and Ethics in Machine Learning|
|Day 6, Tuesday 24 May|
|16:00 - 18:00||Programming: Putting models in production (Part II)|
|Day 7, Tuesday 31 May|
|16:00 - 18:00||New advances in Machine Learning|
|Day 8, Tuesday 7 June|
|16:00 - 18:00||Wrap-up & Case study on ethics and fairness|