Crafting Superior ML Pipelines with SapientML – Kosaku Kimura, Fujitsu
Crafting Superior ML Pipelines with SapientML – Kosaku Kimura, Fujitsu
Introducing SapientML, an emerging project under LF AI&Data, set to transform AutoML. Unlike the other AutoML tools, SapientML goes beyond parameter tuning and algorithm selection by automating AI model creation through direct ML source code generation. This enables quick and accurate optimization of ML pipelines for predictive tasks on tabular data.
SapientML’s unique approach sets it apart. Rather than using search-based tactics found in many AutoML tools, SapientML designs multiple pipelines with the top three ML models and preprocessing components for a given task. It then selects the best one after evaluation. The result is a high-quality ML pipeline with accompanying source code, ready for integration or refinement, in just minutes.
These generated pipelines and ML models are inherently portable and built on well-known libraries. In this session, we will introduce the functionality of SapientML, showcasing its efficiency, transparency, and quality through a live demo. We will also discuss how SapientML enhances AI/MLOps, especially by providing ready-to-use source code.
by The Linux Foundation
linux foundation
