Augment & Valuate : A Data Enhancement Pipeline for Data-Centric AI

TL;DR

This paper introduces a domain-agnostic data enhancement pipeline for image classification that combines valuation, cleansing, and augmentation to improve data quality and model performance, demonstrating competitive results in a data-centric AI competition.

Contribution

It proposes a scalable, generalized data refinement pipeline integrating valuation, cleansing, and augmentation for data-centric AI applications.

Findings

Achieved 84.711% test accuracy in a competitive setting.

Pipeline ranked #6 with Honorable Mention for innovation.

Demonstrated effectiveness of combined data enhancement methods.

Abstract

Data scarcity and noise are important issues in industrial applications of machine learning. However, it is often challenging to devise a scalable and generalized approach to address the fundamental distributional and semantic properties of dataset with black box models. For this reason, data-centric approaches are crucial for the automation of machine learning operation pipeline. In order to serve as the basis for this automation, we suggest a domain-agnostic pipeline for refining the quality of data in image classification problems. This pipeline contains data valuation, cleansing, and augmentation. With an appropriate combination of these methods, we could achieve 84.711% test accuracy (ranked #6, Honorable Mention in the Most Innovative) in the Data-Centric AI competition only with the provided dataset.