Home // ICAS 2024, The Twentieth International Conference on Autonomic and Autonomous Systems // View article

DebiAI: Open-Source Toolkit for Data Analysis, Visualisation and Evaluation in Machine Learning

Authors:
Tom Mansion
Raphaël Braud
Ahmed Amrani
Sabrina Chaouche
Faouzi Adjed
Loïc Cantat

Keywords: Data Analysis; Data Visualization; Bias Detection; Contextual Evaluation; Machine Learning; Trustworthy AI

Abstract:
DebiAI is an open-source tool designed for data analysis, visualization, as well as evaluation and comparison of Machine Learning (ML) models. It is intended to be used both at the stage of the project data preparation, and for the evaluation of the ML models performances. It has a rich and user-friendly graphical interface that allows to visualize, analyze, select, edit and annotate data and metadata, as well as for bias detection and contextual evaluation of ML models. The tool relies on a generic data model, making it applicable to any type of ML task: classification, regression, object detection in images and more. It is an open source code distributed under the Apache License, Version 2.0 . The code is publicly available on https://github.com/debiai and further information along with guidelines for the users can be found on its dedicated website https://debiai.irt-systemx.fr.

Pages: 25 to 30

Copyright: Copyright (c) IARIA, 2024

Publication date: March 10, 2024

Published in: conference

ISSN: 2308-3913

ISBN: 978-1-68558-140-4

Location: Athens, Greece

Dates: from March 10, 2024 to March 14, 2024