CAIML #24
CAIML #24 is going to happen on April 13, 2023, at skillbyte. We will have two talks on machine learning models and machine learning operations, and time for networking with food and drinks
Agenda
18h30 Open Doors
19h00 Welcome & Intro
19h15 Philipp Wendland (PhD Student at Koblenz University of Applied Sciences, Group of Prof. Dr. Maik Kschischo): Neural ODEs - A state-of-the-art Deep Learning approach to process time series data
Neural ODEs are a hybrid deep learning approach based on modelling the dynamic of hidden layers of a neural network in a continuous fashion as an Ordinary Differential Equation (ODE). Due to its continuous nature and promising performance Neural ODEs are a state-of-the-art approach to process (unevenly sampled) multivariate time-series data. Further promising and succesful applications of the Neural ODEs are image classifications, density estimation with continuous normalizing flows and the creation of multi-state survival models. In this talk we want to introduce the general framework of Neural ODEs with a particular focus on its applications to patient data. We will present our extension the so-called Multimodel Neural ODEs to generate highly realistic synthetic patient data based on both static and continuous covariates.
5 Minute Break
19h45 Dr. Andreas Weiden (Co-Lead Cloud / Data Engineering at skillbyte): MLOps pipelines using MLFlow - From training to production
In this talk we explore the world of MLOps pipelines and how MLFlow can be used to facilitate workflows for getting your machine learning models from training to production. We will briefly delve into the tracking aspects of MLFlow and how to store experiments and runs. Next, we will move on to an actual use case that involves managing artefacts generated by multiple training pipelines running on a daily schedule. These artefacts are used in prediction services but also in managed vector search engines such as ElasticSearch and Google VertexAI. A simple microservice that polls the MLFlow registry is used to update both REST-APIs running in Kubernetes and to ingest the models into the vector search services. Finally, we will compare different alternatives that were considered.
20h15 Networking with food and drinks sponsored by skillbyte