A CLUSTER composed of the European projects AEQUITAS, BIAS, FINDHR and MAMMOTH, funded through the Horizon Europe program. This network is a crucial component of the European Commission's strategy to ensure the trustworthiness of AI and it is focused on developing and deploying human-centric, sustainable, secure, inclusive, and trustworthy AI technology.
Mission
What is the cluster’s mission?
The AI Fairness Cluster brings together projects united by a shared mission: identifying and mitigating bias, and making sure AI systems contribute to diversity and inclusion.
Goals
What are the cluster goals?
members
Who are the cluster members?
AEQUITAS will develop a framework to address and tackle the multiple manifestations of bias and unfairness of AI by proposing a controlled experimentation environment for AI developers.
BIAS will empower the AI and Human Resources Management (HRM) communities by addressing and mitigating algorithmic biases.
FINDHR will facilitate the prevention, detection, and management of discrimination in algorithmic hiring and closely related areas involving human recommendation.
Highlights
Check our projects’ highlights
BIAS Co-creation Workshop | Business Modelling for Trustworthy AI Recruitment
Join us for the BIAS online co-creation workshop, "Business Modelling for Trustworthy AI-Powered Recruitment Platforms," on October 29 from 10:00–12:00 CET. This interactive event brings together diverse participants to collaboratively develop a SWOT analysis, a crucial element for a sustainable AI recruitment business model. The workshop will cover topics like the EU AI Act and include breakout sessions on stakeholder mapping and SWOT analysis. Hosted online via MS Teams.
AIMMES 2024 Proceedings
The proceedings from the 1st Workshop on AI Bias: Measurements, Mitigation, Explanation Strategies, co-located with the AI Fairness Cluster Inaugural Conference 2024 in Amsterdam on March 20, 2024, are now available! Check them out today.
BIAS Data Donation Campaign
Champion diversity and innovation in your company: make a difference with your cover letters and CVs! Until 31 January 2025, share anonymized cover letter and CV samples and become our next case study.
AI Fairness Cluster Inaugural Conference & AIMMES'24
The AI Fairness Cluster Inaugural Conference on March 19, 2024, in Amsterdam, brings together experts and researchers from 50+ institutions to address discrimination risks in AI, featuring keynotes, project presentations, and panels, while the co-located AIMMES'24 workshop on March 20 explores technical aspects of measuring, mitigating, and explaining AI biases.
AI Fairness Cluster Flyer
Explore the captivating world of the AI Fairness Cluster through our flyer. Uncover the mission, goals, and vibrant community of members shaping the future of artificial intelligence. Dive into a journey of discovery!
BIAS Trustworthy AI Helix
Join the Trustworthy AI Helix within the BIAS community for project updates and collaboration with experts in Trustworthy AI. This community is essential for disseminating and exploiting BIAS project results, creating a robust and self-sustaining network. External stakeholders are invited to join and contribute to this thriving community.
MAMMOth's FairBench library
ΜΑΜΜΟth's FairBench Python library offers comprehensive AI fairness exploration, generating reports, stamps, and supporting multivalue multiattribute assessments. It seamlessly integrates with numpy, pandas, tensorflow, and pytorch for efficient ML exploration.
BIAS National Labs
BIAS is creating stakeholder communities in two key ecosystems: one focused on HR and recruitment policies, and the other comprising AI experts. Participants will be engaged in various BIAS activities, have access to personalized project information, and play an active role. If interested, please apply to join this initiative.
MAMMOth's FLAC approach for fairness-aware representation learning
MAMMOth's FLAC minimizes mutual information between model features and a protected attribute, using a sampling strategy to highlight underrepresented samples. It frames fair representation learning as a probability matching problem with a bias-capturing classifier, theoretically ensuring independence from protected attributes.
