ralf: Real-time, Accuracy Aware Feature Store Maintenance

Tuesday, October 11, 6:55pm - 7:05pm (EDT)

Time shown in-04:00 America, New York

Speaker: Sarah Wooders, UC Berkeley RISE

Feature stores are becoming ubiquitous in real-time model serving systems, however there has been limited work in understanding how features should be maintained over changing data. In this talk, we present ongoing research at the RISELab on streaming feature maintenance that optimizes both resource costs and downstream model accuracy. We introduce a notion of feature store regret to evaluate feature quality of different maintenance policies, and test various policies on real-world time-series data.

Add to Calendar 2022/10/11 15:55:00 2022/10/11 16:05:00 America/Los_Angeles ralf: Real-time, Accuracy Aware Feature Store Maintenance Speaker: Sarah Wooders, UC Berkeley RISE

Feature stores are becoming ubiquitous in real-time model serving systems, however there has been limited work in understanding how features should be maintained over changing data. In this talk, we present ongoing research at the RISELab on streaming feature maintenance that optimizes both resource costs and downstream model accuracy. We introduce a notion of feature store regret to evaluate feature quality of different maintenance policies, and test various policies on real-world time-series data. Online false MM/DD/YYYY 30 OPAQUE apQfLtmRnzOiFbgoNmal132273

Online

Feature Store Summit 2022

Link: