Picture for Ivan Y. Tyukin

Ivan Y. Tyukin

Stealth edits for provably fixing or attacking large language models

Add code
Jun 18, 2024
Figure 1 for Stealth edits for provably fixing or attacking large language models
Figure 2 for Stealth edits for provably fixing or attacking large language models
Figure 3 for Stealth edits for provably fixing or attacking large language models
Figure 4 for Stealth edits for provably fixing or attacking large language models
Viaarxiv icon

Weakly Supervised Learners for Correction of AI Errors with Provable Performance Guarantees

Add code
Feb 06, 2024
Viaarxiv icon

The Boundaries of Verifiable Accuracy, Robustness, and Generalisation in Deep Learning

Add code
Sep 13, 2023
Viaarxiv icon

How adversarial attacks can disrupt seemingly stable accurate classifiers

Add code
Sep 07, 2023
Figure 1 for How adversarial attacks can disrupt seemingly stable accurate classifiers
Figure 2 for How adversarial attacks can disrupt seemingly stable accurate classifiers
Figure 3 for How adversarial attacks can disrupt seemingly stable accurate classifiers
Figure 4 for How adversarial attacks can disrupt seemingly stable accurate classifiers
Viaarxiv icon

Agile gesture recognition for capacitive sensing devices: adapting on-the-job

Add code
May 12, 2023
Viaarxiv icon

Towards a mathematical understanding of learning from few examples with nonlinear feature maps

Add code
Nov 07, 2022
Viaarxiv icon

Learning from few examples with nonlinear feature maps

Add code
Mar 31, 2022
Figure 1 for Learning from few examples with nonlinear feature maps
Viaarxiv icon

Quasi-orthogonality and intrinsic dimensions as measures of learning and generalisation

Add code
Mar 30, 2022
Figure 1 for Quasi-orthogonality and intrinsic dimensions as measures of learning and generalisation
Figure 2 for Quasi-orthogonality and intrinsic dimensions as measures of learning and generalisation
Figure 3 for Quasi-orthogonality and intrinsic dimensions as measures of learning and generalisation
Figure 4 for Quasi-orthogonality and intrinsic dimensions as measures of learning and generalisation
Viaarxiv icon

Learning from scarce information: using synthetic data to classify Roman fine ware pottery

Add code
Jul 03, 2021
Figure 1 for Learning from scarce information: using synthetic data to classify Roman fine ware pottery
Figure 2 for Learning from scarce information: using synthetic data to classify Roman fine ware pottery
Figure 3 for Learning from scarce information: using synthetic data to classify Roman fine ware pottery
Figure 4 for Learning from scarce information: using synthetic data to classify Roman fine ware pottery
Viaarxiv icon

High-dimensional separability for one- and few-shot learning

Add code
Jun 28, 2021
Figure 1 for High-dimensional separability for one- and few-shot learning
Figure 2 for High-dimensional separability for one- and few-shot learning
Figure 3 for High-dimensional separability for one- and few-shot learning
Figure 4 for High-dimensional separability for one- and few-shot learning
Viaarxiv icon