We've developed deep neural rejection (DNR), a mechanism to detect adversarial examples. It rejects samples that exhibit anomalous feature representations at different network layers. Paper: arxiv.org/pdf/1910.00470 Code will be released as part of secml: secml.gitlab.io

Deep Neural Rejection against Adversarial Examples deepai.org/publication/de… by Angelo Sotgiu et al. including Battista Biggio #NeuralNetwork #ComputerScience

Can Domain Knowledge Alleviate Adversarial Attacks in Multi-Label Classifiers? We investigate how injecting domain-knowledge constraints (first-order logic relationships among classes) into the learning process can help detect adversarial examples. arxiv.org/pdf/2006.03833…

It's with enormous pleasure that I tell you that our last paper has been accepted for publication in the IEEE TPAMI, the most important Machine Learning journal. Check it out! ieeexplore.ieee.org/document/96614… Battista Biggio Stefano Melacci Angelo Sotgiu Marco Gori Ambra

📢We are proud to announce that our paper "Indicators of Attack Failure: Debugging and Improving Optimization of Adversarial Examples " has been accepted to #Neurips2022! w/ Battista Biggio, Luca Demetrio, Ambra, Angelo Sotgiu, Nicholas Carlini, and Fabio Roli 1/n

We’re excited to announce this year’s competitions for SaTML Conference! 🎉 Get ready for four fun challenges tackling prompt injection, data leakage, membership inference, and malware detection. satml.org/competitions/ Let’s dive in! 🧵1/5

Exciting news!🔥 sAIfer Lab is organizing a challenge on Robust Android Malware Detection! Hosted by IEEE #SaTML2025, it aims to evaluate #machinelearning methods in the context of malicious attacks against the #Android systems. ramd-competition.github.io Deadline: March 31st

📢 Check out the S2RAI Track (at SAC26 Conference)! I am co-chairing with Simon Burton, Marta Catillo, Jeremie Guiochet, and Tommaso Zoppi 🗓️ Deadline: September 26th, 2025 ℹ️ Info here: sites.google.com/view/s2rai-sac/ 🇬🇷 SAC26 Website: sigapp.org/sac/sac2026/

🆕 UPDATE: the deadline has been extended to October 10, 2025 (EDT)!

🆕 There's still time to submit your work! The deadline has been extended to October 17, 2025 (EDT)!