5 papers
The mathematics of adversarial attacks in AI -- Why deep learning is unstable despite the existence of stable neural networks
Alexander Bastounis, Anders C Hansen, Verner VlaÄiÄ
The unprecedented success of deep learning (DL) makes it unchallenged when it comes to classification problems. However, it is well established that the current DL methodology prod…
The Boundaries of Verifiable Accuracy, Robustness, and Generalisation in Deep Learning
Alexander Bastounis, Alexander N. Gorban, Anders C. Hansen +5
In this work, we assess the theoretical limitations of determining guaranteed stability and accuracy of neural networks in classification tasks. We consider classical distribution-…
Stealth edits to large language models
Oliver J. Sutton, Qinghua Zhou, Wei Wang +4
We reveal the theoretical foundations of techniques for editing large language models, and present new methods which can do so without requiring retraining. Our theoretical insight…
How adversarial attacks can disrupt seemingly stable accurate classifiers
Oliver J. Sutton, Qinghua Zhou, Ivan Y. Tyukin +3
Adversarial attacks dramatically change the output of an otherwise accurate learning system using a seemingly inconsequential modification to a piece of input data. Paradoxically,…
On the consistent reasoning paradox of intelligence and optimal trust in AI: The power of 'I don't know'
Alexander Bastounis, Paolo Campodonico, Mihaela van der Schaar +2
We introduce the Consistent Reasoning Paradox (CRP). Consistent reasoning, which lies at the core of human intelligence, is the ability to handle tasks that are equivalent, yet des…