F-IVM: Learning over Fast-Evolving Relational Data

TL;DR

F-IVM is a system designed for real-time analytics on rapidly changing relational data, enabling machine learning tasks like model selection, Chow-Liu trees, and ridge linear regression.

Contribution

It introduces a system that efficiently supports real-time machine learning over fast-evolving relational datasets, a novel capability in data analytics.

Findings

Demonstrates real-time analytics for multiple ML applications

Supports fast updates in relational databases for ML tasks

Enables efficient model selection and regression on dynamic data

Abstract

F-IVM is a system for real-time analytics such as machine learning applications over training datasets defined by queries over fast-evolving relational databases. We will demonstrate F-IVM for three such applications: model selection, Chow-Liu trees, and ridge linear regression.