Home // International Journal On Advances in Systems and Measurements, volume 17, numbers 3 and 4, 2024 // View article
Authors:
Tom Mansion
Raphaël Braud
Faouzi Adjed
Ahmed Amrani
Sabrina Chaouche
Fady Bekkar
Yoann Randon
Martin Gonzalez
Loïc Cantat
Keywords: Data Analysis; Data Visualization; Bias Detection; Human-Centered Machine Learning; Trustworthy AI.
Abstract:
We present DebiAI, a powerful open source tool crafted to streamline data analysis, visualization, and the comprehensive evaluation and comparison of Machine Learning models. It serves as a versatile companion throughout the entire machine learning workflow, from project data preparation to model performance assessment.With its intuitive and feature-rich graphical interface, DebiAI enables users to effortlessly visualize, explore, select, edit, and annotate both data and metadata. The tool is also equipped for bias detection and contextual evaluation of ML models, ensuring a thorough and fair analysis. Built on a flexible, generic data model, DebiAI is adaptable to a wide range of ML tasks, including classification, regression, and object detection in images, as well as a variety of tasks for time-series and more. Released under the Apache License, Version 2.0, it offers an accessible and linearly scalable solution for ML practitioners of all levels. The code for the proposed tool is publicly available at https://github.com/debiai; and other information and user guidelines are available on the dedicated website: https://debiai.irt-systemx.fr.
Pages: 83 to 95
Copyright: Copyright (c) to authors, 2024. Used with permission.
Publication date: December 30, 2024
Published in: journal
ISSN: 1942-261x