📰 Calling all journalists and fact-checkers!
Be part of a European pilot to improve AI tools that help detect and analyse disinformation.
Your feedback will guide the next generation of trustworthy, ethical AI for media.
Sign up now 👉 forms.gle/S6RvKp91u5hQ...
#AI4TRUST #HorizonEU
New handbook published by @ebu.ch - the title sums it up: "Handbook on the legal and ethical obligations to developers and deployers of AI-based fact-checking tools".
Available for download.
Image with details about the event as indicated here and on website
Join us on 24 June from 2-5 pm CET for our second online webinar in which we present outcomes of our work on #disinformation detection and content analysis. Focus: research. More, incl. registration link: www.veraai.eu/posts/two-ve...
CC @sympap.bsky.social @ivansrba.bsky.social
Alert: 📌 Registration open to join the MediaEval 2025 challenge.
📂 Data & Submission Instructions
Available in the official GitHub repository:
👉 github.com/mever-team/m...
📅 Important Dates
Data release: June 20
Runs due: September 15
Paper submission: October 8
Workshop: October 25–26
Website 👇
📢 ELLIOT is coming! A €25M #HorizonEurope project to develop open, trustworthy Multimodal Generalist Foundation Models, #MGFM, for real-world applications. Starting July, it brings 30 partners from 12 countries to shape Europe’s #AI future.
🔍 Follow for updates on #OpenScience & #FoundationModels.
Just over one week to go before our first workshop takes place in which we present outcomes of our work. On 17 & 24 June, 2-5 pm we invite the #verification, #factchecking, #disinformation detection community to meet us on Zoom.
Registration required. Hope to see you then!
We're inviting the #verification, #factchecking and #disinformation detection community to 2 workshops in which we showcase veraAI results: 17 June is targeting #journalists and #factcheckers; 24 June focusses on the R&D community. Both events run from 2-5 pm. Registration required. See you soon!? 😃
veraAI partner @disinfo.eu / Ana Romero-Vicente (@anicanaca.bsky.social) authored a report entitled "Visual assessment of Coordinated Inauthentic Behaviour in disinformation campaigns". Editor: @netosessa.bsky.social. We provide a summary and the full publication here. www.veraai.eu/posts/report...
Today, we are publishing the first-ever International AI Safety Report, backed by 30 countries and the OECD, UN, and EU.
It summarises the state of the science on AI capabilities and risks, and how to mitigate those risks. 🧵
Full Report: assets.publishing.service.gov.uk/media/679a0c...
1/21
📢 We will be co-organizing the 4th edition of the Multimedia AI against Disinformation (MAD'25) workshop on June 30 @ Chicago, USA.
👉 more information on topics of interest, dates and submissions: mad2025.aimultimedialab.ro
ℹ️ The workshop is supported by @vera-ai.bsky.social.
Hello Bluesky! We are MedDMO, a regional hub of @edmo-eu.bsky.social covering Greece, Cyprus, and Malta. We bring together research, fact-checking, and media organizations that conduct internationally acknowledged research and activities in the area of disinformation.
Let's connect!
Not to mention the complexities of localizing images captured in non-urban areas that are mostly featuring vegetation, soil, and other kinds of "generic" natural scenes.
🌍 Guessing where an image was taken is a hard, and often ambiguous problem. Introducing diffusion-based geolocation—we predict global locations by refining random guesses into trajectories across the Earth's surface!
🗺️ Paper, code, and demo: nicolas-dufour.github.io/plonk
More details in our Future Generation Computer Systems article "Reducing inference energy consumption using dual complementary CNNs": www.sciencedirect.com/science/arti...
Building on this concept, we manage to build inference systems based on highly complementary pairs of neural networks that strike very good trade-off between accuracy and energy consumption. ...
Energy consumption of AI becomes a priority as such systems are widely deployed and are responsible for a big part of society's energy footprint. In our latest work, we proposed the concept of neural network "complementarity" to quantify the extent that two NNs lead to complementary predictions...
Book outline
Over the past decade, embeddings — numerical representations of machine learning features used as input to deep learning models — have become a foundational data structure in industrial machine learning systems. TF-IDF, PCA, and one-hot encoding have always been key tools in machine learning systems as ways to compress and make sense of large amounts of textual data. However, traditional approaches were limited in the amount of context they could reason about with increasing amounts of data. As the volume, velocity, and variety of data captured by modern applications has exploded, creating approaches specifically tailored to scale has become increasingly important. Google’s Word2Vec paper made an important step in moving from simple statistical representations to semantic meaning of words. The subsequent rise of the Transformer architecture and transfer learning, as well as the latest surge in generative methods has enabled the growth of embeddings as a foundational machine learning data structure. This survey paper aims to provide a deep dive into what embeddings are, their history, and usage patterns in industry.
Cover image
Just realized BlueSky allows sharing valuable stuff cause it doesn't punish links. 🤩
Let's start with "What are embeddings" by @vickiboykis.com
The book is a great summary of embeddings, from history to modern approaches.
The best part: it's free.
Link: vickiboykis.com/what_are_emb...
The Cosmos suite of neural tokenizers for images & videos is impressive.
Cosmos is trained on diverse high-res imgs & long-vids, scales well for both discrete & continuous tokens, generalizes to multiple domains (robotics, driving, egocentric ...) & has excellent runtime
github.com/NVIDIA/Cosmo...
