Konrad Rieck 🌈

Congratulations and welcome in the system!

🚨 Got a great idea for an AI + Security competition?

@satml.org is now accepting proposals for its Competition Track! Showcase your challenge and engage the community.

👉 satml.org/call-for-com...
🗓️ Deadline: Aug 6

XAI 2025: Best Paper Award Congratulations to BIFOLD researchers Simon Letzgus, Klaus-Robert Müller and Grégoire Montavon, whose publication: XpertAI: uncovering regression model strategies for sub-manifolds&nbsp...

🎉 Congratulations to BIFOLD researchers Simon Letzgus, Klaus-Robert Müller, and Grégoire Montavon on receiving the Best Paper Award at the 3rd World Conference on eXplainable Artificial Intelligence (hashtag#XAI2025).

www.bifold.berlin/news-events/...

@tuberlin.bsky.social

This work emerged from a spontaneous collaboration with the group of @matthiasboehm7.bsky.social at our institute
@bifold.berlin and CASA.

If you'd like to learn more, check out our paper: mlsec.org/docs/2025-ic...

Code for crafting your own Chimera examples will follow soon.

4/4

Fortunately, exploiting these discrepancies is not straightforward, as real-world inputs are typically limited in precision. Still, with some effort, we manage to induce divergent predictions even from 8-bit integers like image pixels, resulting in Chimera images 👾.

3/4

Visualization of a matrix mutplication. The matrices are decomposed into blocks that are processed individually.

Technically, we build on the non-associativity of floating-point arithmetic. When computing convolutions or matrix multiplications, the backends split data into blocks and process them in different orders, introducing slight deviations and exposing an attack surface.

2/4

Visualization of the attack. A dog is shown for which three backends yield the prediction "boxer", while one backend returns "castle".

Today, Jonas presents a new type of adversarial examples at
@icmlconf.bsky.social!

We exploit subtle numerical differences between linear algebra backends and craft inputs that yield different predictions from the same model depending on the backend used 🤯 mlsec.org/docs/2025-ic...

1/4

eurips.cc A NeurIPS-endorsed conference in Europe held in Copenhagen, Denmark

NeurIPS is endorsing EurIPS, an independently-organized meeting which will offer researchers an opportunity to additionally present NeurIPS work in Europe concurrently with NeurIPS.

Read more in our blog post and on the EurIPS website:
blog.neurips.cc/2025/07/16/n...
eurips.cc

Call for Competitions Competition proposal deadline: August 6, 2025 Decision notification: August 27, 2025

We’re happy to announce the Call for Competitions for
@satml.org

The competition track has been a highlight of SaTML, featuring exciting topics and strong participation. If you’d like to host one for SaTML 2026, visit:

👉 satml.org/call-for-com...
⏰ Deadline: Aug 6

Homemade pizza

Homemade pizza

Practicing for my plan B career

NebulaStream Goes Open Source NebulaStream, an extensible, high-performance streaming engine for multi-modal edge applications is now open-source. Developed for the Internet of Things by BIFOLD, DIMA, and DFKI researchers, this no...

🚀 NebulaStream is now Open Source!

NebulaStream is an Extensible, High-Performance Streaming Engine for Multi-Modal Edge Applications. This novel stream processing engine analyzes sensor data in real time. Now freely available under the Apache 2.0 license.

www.bifold.berlin/news-events/...

IEEE Conference on Secure and Trustworthy Machine Learning (SaTML), March 23-25, 2025, Munich Submission deadline: September 24, 2025

We're excited to announce the Call for Papers for SaTML 2026, the premier conference on secure and trustworthy machine learning @satml.org

We seek papers on secure, private, and fair learning algorithms and systems.

👉 satml.org/call-for-pap...
⏰ Deadline: Sept 24

Thank you senator Katharina Günther-Wünsch, Senatsverwaltung für Bildung, Jugend und Familie, for taking the time to visit the BIFOLD booth at LNdW 2025. We believe that AI education belongs in schools and we try to get the next generation of scientists interested in AI research at an early stage. #KImachtschule

Thank you senator Katharina Günther-Wünsch, Senatsverwaltung für Bildung, Jugend und Familie, for taking the time to visit the BIFOLD booth at LNdW 2025. We
believe that AI education belongs in schools and we try to get the next generation of scientists interested in AI research at an early stage.

Thrilled to share a recent opinion piece at the IEEE Security and Privacy (Vol. 23, Issue 3).

Basically a long-term perspective on the field meant for both researchers and practitioners.

📝 ieeexplore.ieee.org/stamp/stamp....

Our demo "mlidea: Interactively Improving ML Data Preparation Code via 'Shadow Pipelines'" was accepted at VLDB! 🥳

We demo suggestions for ML pipelines, similar to IntelliJ code inspections or Grammarly suggestions

youtu.be/ePGm1J6S2qk

Joint work w/ @mersault.bsky.social @p-groth.bsky.social

All @acm.org publications will be 100% Open Access as of January 2026. When we announced this at POPL and CHI this year, conference participants spontaneously erupted in applause. The CS community is excited about ACM's move to OA!

Table with three columns: one "other approaches", all with 0%; one "our approach", with bold, non-zero percentages; and one "improvement", showing infinite improvement in all cases.

Occasionally, I get a reviewer complaining, “You don’t compare against any other approach. So how can we know you’re better?” So I found a simple way to still add the "bold number column" that apparently everybody wants these days.

Huge congrats to @spalab.cs.ucr.edu's Georgi Ganev for receiving the Distinguished Paper Award at IEEE S&P for his work "The Inadequacy of Similarity-based Privacy Metrics: Privacy Attacks against “Truly Anonymous” Synthetic Datasets."

Paper: arxiv.org/pdf/2312.051...

Haven't seen this on Bluesky yet: S&P 2027 will take place in Montreal, Canada!

🎉Hats off to Dr. Stefan Blücher! He defended his PhD yesterday🎓

Thesis: "Towards Scalable and Transparent ML Algorithms with Applications in Explainable AI and Quantum Chemistry"

Thanks to all supporters!
#XAI #QuantumML #MLResearch #PhDCompleted @tuberlin.bsky.social @fraunhoferhhi.bsky.social

Bid to host SaTML 2026 Thank you for considering to host SaTML! SaTML has been organized as a 3 day conference so far. We are looking for volunteers interested in finding a venue to host the conference in 2026. By submitti...

🌍 Help shape the future of SaTML!

We are on the hunt for a 2026 host city - and you could lead the way. Submit a bid to become General Chair of the conference:

forms.gle/vozsaXjCoPzc...

🚀 Exciting Announcement! 🚀

Get ready for the 18th ACM Workshop on Artificial Intelligence and Security (AISec 2025)!

📍Co-located: ACM CCS www.sigsac.org/ccs/CCS2025/

🗓️ Deadline: June 20th, 2025
🌐 Website: aisec.cc

w/
Ruoxi Jia and Matthew Jagielski

The Spring Reception at #BIFOLDDay 2025 is officially open!

Thanks to Senator Dr. Ina Czyborra and Dr. Tina Klüwer (BMBF) for opening remarks and the Keynote by Prof. Dr. Matthias Bethge on “Cosmopolitan AI”.
Excited to celebrate 2024 and look ahead!

@regberlin.bsky.social @tuberlin.bsky.social

Thank you to Stefan Halfpap, Haralampos Gavriilidis, Lukas Pirch, and Stephanie Brandl for their excellent tutorials this afternoon at #BIFOLDDay2025.
We look forward to the Spring Reception starting at 4 PM!

@lpirch.bsky.social @stephaniebrandl.bsky.social @harrygav.bsky.social

Tomorrow is BIFOLD Day (April 30, 2025). We are looking forward to Tutorials on Data Visualization, Efficient Data Loading, Machine Learning with PyTorch-Geometric and Natural Language Processing.

A conference poster at ICLR 2025 titled “Fast Jacobians and Hessians by leveraging sparsity”.

If you are at #ICLR2025 and want to chat about automatic sparse differentiation (or just grab a sticker), come see me at poster 471!

Our paper "Top Score on the Wrong Exam" paper will be presented at #ISSTA25 🐣 in Trondheim!

📝https://mpi-softsec.github.io/papers/ISSTA25-topscore.pdf
🧑‍💻https://github.com/niklasrisse/TopScoreWrongExam

// @nrisse.bsky.social @fuzzing.bsky.social

I had two NSF grants terminated last week, for a total of just under $1M. The largest one had just started and its goal was to provide users with more context and information on the content that they see on social media.

I was invited to review for the "Journal of Pipeline Systems Engineering and Practice", seems our work on ML pipelines is finally recognised by other communities as well ;D

Bid to host SaTML 2026 Thank you for considering to host SaTML! SaTML has been organized as a 3 day conference so far. We are looking for volunteers interested in finding a venue to host the conference in 2026. By submitti...

🚨 SaTML is searching for its 2026 home!
Interested in becoming General Chair and hosting the conference in your city or institution? We’d love to hear from you. Place a bid here:
👉 forms.gle/kbxtwZddpcLD...