Description
This is a free event!
Presenter: Dr Leonardo Ramirez-Lopez, Head of Data Science at BÜCHI Labortechnik AG
Title: Why refit when you can retrieve? Introducing liblex, an algorithmic framework for local model retrieval in infrared spectroscopy
Abstract
Local chemometric models (e.g. LOCAL) predict well on complex soil spectra, but they refit a model for every sample and need the full training data at prediction time.
In this webinar, I present liblex, an algorithmic framework that replaces per-query refitting with retrieval from a library of local models. I test it in near and medium infrared spectroscopy with three soil applications, including a cross-continental transfer, and discuss interpretability, uncertainty and deployment without sharing training data. The framework is available in "resemble" an open-source R package.
Tickets for good, not greed Humanitix dedicates 100% of profits from booking fees to charity
