The Principles of Data-Centric AI (DCAI)

TL;DR

This paper introduces Data-Centric AI (DCAI), emphasizing the importance of data quality over models, and outlines six guiding principles to advance AI systems through a data-focused approach.

Contribution

It provides one of the first comprehensive overviews of DCAI, formulating six principles to guide research and practice in data-centric AI development.

Findings

Highlights the shift from model-centric to data-centric AI

Proposes six guiding principles for DCAI

Outlines future directions for DCAI research and practice

Abstract

Data is a crucial infrastructure to how artificial intelligence (AI) systems learn. However, these systems to date have been largely model-centric, putting a premium on the model at the expense of the data quality. Data quality issues beset the performance of AI systems, particularly in downstream deployments and in real-world applications. Data-centric AI (DCAI) as an emerging concept brings data, its quality and its dynamism to the forefront in considerations of AI systems through an iterative and systematic approach. As one of the first overviews, this article brings together data-centric perspectives and concepts to outline the foundations of DCAI. It specifically formulates six guiding principles for researchers and practitioners and gives direction for future advancement of DCAI.