Enhancing Accessible, Ethical and Unbiased AI. Funded by @HorizonEU under GA 101177912
- Tweets 124
- Following 81
- Followers 77
- Likes 48
🇪🇺 The Digital Omnibus revises EU AI Act implementation: simpler compliance, adjusted high-risk AI rules, and stronger safeguards against harmful AI uses.
How can Europe balance innovation and fundamental rights?
Read @profpdh’s analysis for #ALFIE:
alfie-project.eu/eu-ai-act-i…
1
1
40
🌍 ALFIE partners from UAB and Catalink joined IEEE CBMI 2026 in Toulouse, sharing research on ethical and inclusive AI.
From gender and cultural diversity in AI datasets to accessibility and fairness, the focus was clear: AI must work for everyone.
#ALFIEProject #CBMI2026
1
1
1
57
🔍 Three key research topics:
🤟 Accessible sign language avatars
🚗 Demographic bias in driver drowsiness detection
🎙️ Intersectional bias in speech recognition
Together, they support ALFIE’s mission to make AI fairer and more inclusive.
🔗 alfie-project.eu/from-ethica…
#AI
1
15
1/2 🔄 Cluster-Based Drift Detection for Efficient AI Adaptation
Does AI really need constant retraining as data changes?
Researchers contributing to the #ALFIEProject propose a cluster-based approach to detect changes and retrain models only when necessary.
1
1
1
63
2/2 📊 The results?
Across 5 datasets and 6 adaptation strategies, the approach reduced retraining effort by up to 75% while maintaining competitive predictive performance.
🔗 Read the study:
doi.org/10.1016/j.eswa.2026.…
#AIResearch #MachineLearning #AI #ML #Data #Training
1
1
20
Making AI easier to build and understand!
In our latest blog, Subhaditya Mukherjee from TU/e explores how the #ALFIEProject AutoML Engine makes AI development more accessible and responsible.
🔗 Read the full article:
alfie-project.eu/building-th…
1
1
1
53
The ALFIE AutoML Engine brings together:
🔹 Automated classification for text, audio & images
🔹 Data quality checks & bias detection
🔹 Explainable AI (XAI)
🔹 Ethical and responsible AI
🔹 An AI-powered assistant to guide users
#AutoML #ResponsibleAI #AI
1
1
43
A model may stay the same, but the world around it does not. An ALFIE review of 147 studies explores four key challenges over time: data drift, concept drift, skewed learning, and catastrophic forgetting.
1
2
4
75
The real question is not only whether a model works at deployment, but whether it stays reliable as conditions change. For ALFIE, this is central to lifecycle monitoring under the EU AI Act.
Full Article : tandfonline.com/doi/full/10.…
#ALFIE #AI #EUAIAct
1
3
62
On 25 September, Reyhaneh Sohrabi from the ALFIE team at UAB presented “AI Listens, But Not to Everyone” at Researchers’ Night 2026 in Barcelona.
The talk explored the growing use of Automatic Speech Recognition in real-world settings and its wider societal impact.
ALT Reyhaneh Sohrabi is standing next to the Researchers' Night booth
ALT Reyhaneh Sohrabi is presenting in researchers' night event.
ALT Researcher's night event.
1
3
65
It highlighted how recognition errors can affect low-resource language speakers, older people, people with speech impairments, and accented English speakers differently.
These challenges are central to ALFIE’s work on inclusive and ethical AI.
#ALFIE #ASR #EthicalAI
1
41
A city can be more digital, connected, and automated, and still leave people behind.
At #DIGITALUNGA, more than 350 participants from over 50 countries joined the conversation on what human-centred cities really require: accessibility, trust, and meaningful human choice.
1
28
For #ALFIEProject, this is exactly why inclusive AI matters.
It is not enough to ask whether a system works.
We also need to ask: Who can access it? Who can question it? And who might still be left out?
#HumanCentredAI #InclusiveAI
1
8
What does a human-centric city look like in the age of AI? 🌍🤖
Join @ITU & @TransmediaCat (UAB) on Sept 21 for a #DigitalUNGA side event on AI, smart cities, and digital inclusion.
📅 21 Sept 2026 | 💻 Online 🔗 🔗webs.uab.cat/transmedia/digi…
#SmartCities #AI #Accessibility
37
Back from summer break & heading straight to Brussels for the #OPENVERSE Summit! 🌍
Excited to highlight Prof. Pilar Orero speaking on Ethical Virtual Worlds:
📅 Thu, 10 Sept | 16:45
📍 Brussels
🔗 open-verse.eu/event/openvers…
32
An AI researcher, a geneticist, and an architect looking at the same dataset will rarely see the same problem. 🧵📷
1/ That diversity was the core of the Una Europa Challenge. Participants from different backgrounds came together to learn about LLMs, test models, and explore bias
ALT A group of UNA Europa participants working together at their computers
ALT Four panelist address a room full of participants in the UNA Europa Challenge
1
26
4/ Participants also put the ALFIE platform to the test. Building ethical AI requires perspectives from people who understand data, society, and human behavior differently—spotting risks AI specialists might overlook.
1
12
5/ Huge thanks to Universidad Complutense de Madrid (@ucm) for hosting this incredible interdisciplinary experience!
13