DiffML: End-to-end Differentiable ML Pipelines

TL;DR

DiffML introduces a novel approach to constructing end-to-end differentiable machine learning pipelines, enabling joint training of data preprocessing and modeling steps through backpropagation, which opens new research avenues.

Contribution

This paper proposes the concept of differentiable ML pipelines, demonstrating how typical preprocessing steps can be formulated as differentiable programs for joint optimization.

Findings

Initial ideas for differentiable data cleaning and feature selection

Feasibility of joint training of preprocessing and models

Discussion of research challenges and future directions

Abstract

In this paper, we present our vision of differentiable ML pipelines called DiffML to automate the construction of ML pipelines in an end-to-end fashion. The idea is that DiffML allows to jointly train not just the ML model itself but also the entire pipeline including data preprocessing steps, e.g., data cleaning, feature selection, etc. Our core idea is to formulate all pipeline steps in a differentiable way such that the entire pipeline can be trained using backpropagation. However, this is a non-trivial problem and opens up many new research questions. To show the feasibility of this direction, we demonstrate initial ideas and a general principle of how typical preprocessing steps such as data cleaning, feature selection and dataset selection can be formulated as differentiable programs and jointly learned with the ML model. Moreover, we discuss a research roadmap and core challenges that have to be systematically tackled to enable fully differentiable ML pipelines.