AlphaD3M: Machine Learning Pipeline Synthesis

TL;DR

AlphaD3M is an AutoML system that uses meta reinforcement learning and sequence models to automatically synthesize machine learning pipelines, offering competitive performance, faster computation, and inherent explainability.

Contribution

It introduces AlphaD3M, a novel AutoML approach leveraging meta reinforcement learning with sequence models and self-play for pipeline synthesis.

Findings

AlphaD3M achieves performance comparable to state-of-the-art AutoML systems.

AlphaD3M reduces computation time from hours to minutes.

AlphaD3M provides explainability by design.

Abstract

We introduce AlphaD3M, an automatic machine learning (AutoML) system based on meta reinforcement learning using sequence models with self play. AlphaD3M is based on edit operations performed over machine learning pipeline primitives providing explainability. We compare AlphaD3M with state-of-the-art AutoML systems: Autosklearn, Autostacker, and TPOT, on OpenML datasets. AlphaD3M achieves competitive performance while being an order of magnitude faster, reducing computation time from hours to minutes, and is explainable by design.