Validation in Machine Learning

​Description:
In this talk, we explore key aspects of machine learning model validation, with a focus on practical methodologies and automation. It begins with an overview of several well-known validation tests, highlighting both their similarities and distinctions. Finally, we demonstrate how understanding these similarities and differences enabled us at TBC to automate the validation process as an extension of the end-to-end automated model development pipeline—enhancing both efficiency and consistency.

​Speaker:
Levan is one of the principal contributors to the modeling validation extension for end-to-end AutoML pipeline developed and used by the Risk Modeling Data Science Team at TBC Bank, Georgia. Applying ML techniques to business and scientific research projects, with several years of experience in team leading focused on mathematical simulations of technological processes. This includes both deterministic models (ODEs, PDEs) and data-driven statistical models built with modern ML frameworks. Author and co-author of multiple scientific publications and conference papers.

​Location:
Rooms Hotel Tbilisi, Central Room -1 floor (The main entrance, Chovelidze street side)