Incremental ML for real-time data
10:30 - 10:50, 25th of May (Thursday) 2023/ DEVTRENDS STAGE
Batch processing of models works best when describing processes that change slowly. These times are however nowhere near slow. Monitoring and re-training is an option, but it's outdated at the same time you deploy the model. We're going to evaluate when to consider incremental training instead and adjust the ML model to incoming events ASAP.