ML in Practice - Interaction Between ML Modeling and Downstream Design Decisions

Explore the practical aspects of Machine Learning from an engineering perspective in this insightful talk. Discover why accuracy is often the least interesting aspect of model selection in real-world applications. Learn about a specific project at Ericsson for predicting mobility in telecommunications networks, used as a demonstrator to illustrate the impact of model selection on final tasks. Understand the significant implications that seemingly simple model choices can have on the overall solution. If time permits, delve into a custom wait-free implementation of a sparse graph utilizing approximate updates, developed for this particular use case. Gain valuable insights from Jesper Derehag, Sr. Machine Learning Engineer at Ericsson AB, as he shares his expertise in this 1-hour 27-minute presentation recorded at an Ericsson meetup on December 6, 2022.

ML in Practice - interaction between ML modeling and downstream design decisions