Ilia Shumailov
University of Cambridge
H-index: 17
Europe-United Kingdom
Top articles of Ilia Shumailov
Fairness Feedback Loops: Training on Synthetic Data Amplifies Bias
arXiv preprint arXiv:2403.07857
2024/3/12
Ilia Shumailov
H-Index: 7
Nicolas Papernot
H-Index: 32
Inexact unlearning needs more careful evaluations to avoid a false sense of privacy
arXiv preprint arXiv:2403.01218
2024/3/2
Architectural Neural Backdoors from First Principles
arXiv preprint arXiv:2402.06957
2024/2/10
Buffer Overflow in Mixture of Experts
arXiv preprint arXiv:2402.05526
2024/2/8
Jamie Hayes
H-Index: 11
Ilia Shumailov
H-Index: 7
Beyond Laplace and Gaussian: Exploring the Generalized Gaussian Mechanism for Private Machine Learning
2023/10/13
LLM Censorship: The Problem and its Limitations
2023/10/13
Revisiting block-based quantisation: What is important for sub-8-bit LLM inference?
arXiv preprint arXiv:2310.05079
2023/10/8
Beyond Labeling Oracles: What does it mean to steal ML models?
arXiv preprint arXiv:2310.01959
2023/10/3
Human-Producible Adversarial Examples
arXiv preprint arXiv:2310.00438
2023/9/30
Yue Gao
H-Index: 1
Ilia Shumailov
H-Index: 7
Robert Mullins
H-Index: 13
Ross Anderson
H-Index: 14
Kassem Fawaz
H-Index: 16
SEA: Shareable and Explainable Attribution for Query-based Black-box Attacks
arXiv preprint arXiv:2308.11845
2023/8/23
When the curious abandon honesty: Federated learning is not private
2023/7/3
Roei Schuster
H-Index: 7
Ali Shahin Shamsabadi
H-Index: 8
Ilia Shumailov
H-Index: 7
Nicolas Papernot
H-Index: 32
Reconstructing Individual Data Points in Federated Learning Hardened with Differential Privacy and Secure Aggregation
2023/7/3
Roei Schuster
H-Index: 7
Ali Shahin Shamsabadi
H-Index: 8
Ilia Shumailov
H-Index: 7
Nicolas Papernot
H-Index: 32
Gradients Look Alike: Sensitivity is Often Overestimated in DP-SGD
arXiv preprint arXiv:2307.00310
2023/7/1
When Vision Fails: Text Attacks Against ViT and OCR
arXiv preprint arXiv:2306.07033
2023/6/12
The Curse of Recursion: Training on Generated Data Makes Models Forget
arXiv preprint arXiv:2305.17493
2023/5/27
Boosting Big Brother: Attacking Search Engines with Encodings
2023/10/16
Revisiting Automated Prompting: Are We Actually Doing Better?
arXiv preprint arXiv:2304.03609
2023/4/7
Architectural backdoors in neural networks
2023
Mikel Bober-Irizar
H-Index: 4
Ilia Shumailov
H-Index: 7
Robert Mullins
H-Index: 13
Nicolas Papernot
H-Index: 32
Tubes Among Us: Analog Attack on Automatic Speaker Identification
2023
Rethinking Image-Scaling Attacks: The Interplay Between Vulnerabilities in Machine Learning Systems
2022/6/28
Co-Authors
