86 citations · 286 across the 31 of their papers we have counts for
33 papers
Private online learning and prediction for Littlestone classes
Amartya Sanyal
We study mistake bounds for differentially private online learning and online prediction under oblivious realisable adversaries. Online learning requires the learner to release a h…
Adaptive Sampling and Clipping for Private Worst-Case Group Optimization
Max Cairney-Leeming, Amartya Sanyal, Christoph H. Lampert
A central requirement for the acceptance of machine learning methods for human-centric tasks is that they should be fair, in the sense that they should work comparably well for ind…
LoRA and Privacy: When Random Projections Help (and When They Don't)
Yaxi Hu, Johanna Düngler, Bernhard Schölkopf +1
We introduce the (Wishart) projection mechanism, a randomized map of the form with and study its differential privacy properties. For ve…
Learning in an Echo Chamber: Online Learning with Replay Adversary
Daniil Dmitriev, Harald Eskelund Franck, Carolin Heinzler +1
As machine learning systems increasingly train on self-annotated data, they risk reinforcing errors and becoming echo chambers of their own beliefs. We model this phenomenon by int…
Online Learning and Unlearning
Yaxi Hu, Bernhard Schölkopf, Amartya Sanyal
We formalize the problem of online learning-unlearning, where a model is updated sequentially in an online setting while accommodating unlearning requests between updates. After a…
Differentially Private Steering for Large Language Model Alignment
Anmol Goel, Yaxi Hu, Iryna Gurevych +1
Aligning Large Language Models (LLMs) with human values and away from undesirable behaviors (such as hallucination) has become increasingly important. Recently, steering LLMs towar…
What Makes and Breaks Safety Fine-tuning? A Mechanistic Study
Samyak Jain, Ekdeep Singh Lubana, Kemal Oksuz +4
Safety fine-tuning helps align Large Language Models (LLMs) with human preferences for their safe deployment. To better understand the underlying factors that make models safe via…
Robust Mixture Learning when Outliers Overwhelm Small Groups
Daniil Dmitriev, Rares-Darius Buhai, Stefan Tiegel +5
We study the problem of estimating the means of well-separated mixtures when an adversary may add arbitrary outliers. While strong guarantees are available when the outlier fractio…
Accuracy on the wrong line: On the pitfalls of noisy data for out-of-distribution generalisation
Amartya Sanyal, Yaxi Hu, Yaodong Yu +3
"Accuracy-on-the-line" is a widely observed phenomenon in machine learning, where a model's accuracy on in-distribution (ID) and out-of-distribution (OOD) data is positively correl…
Provable Privacy with Non-Private Pre-Processing
Yaxi Hu, Amartya Sanyal, Bernhard Schölkopf
When analysing Differentially Private (DP) machine learning pipelines, the potential privacy cost of data-dependent pre-processing is frequently overlooked in privacy accounting. I…
On the Growth of Mistakes in Differentially Private Online Learning: A Lower Bound Perspective
Daniil Dmitriev, Kristóf Szabó, Amartya Sanyal
In this paper, we provide lower bounds for Differentially Private (DP) Online Learning algorithms. Our result shows that, for a broad class of -DP online algorithm…
Can semi-supervised learning use all the data effectively? A lower bound perspective
Alexandru Ţifrea, Gizem Yüce, Amartya Sanyal +1
Prior works have shown that semi-supervised learning algorithms can leverage unlabeled data to improve over the labeled sample complexity of supervised learning (SL) algorithms. Ho…
How robust accuracy suffers from certified training with convex relaxations
Piersilvio De Bartolomeis, Jacob Clarysse, Amartya Sanyal +1
Adversarial attacks pose significant threats to deploying state-of-the-art classifiers in safety-critical applications. Two classes of methods have emerged to address this issue: e…
PILLAR: How to make semi-private learning more effective
Francesco Pinto, Yaxi Hu, Fanny Yang +1
In Semi-Supervised Semi-Private (SP) learning, the learner has access to both public unlabelled and private labelled data. We propose a computationally efficient algorithm that, un…
Certifying Ensembles: A General Certification Theory with S-Lipschitzness
Aleksandar Petrov, Francisco Eiras, Amartya Sanyal +2
Improving and guaranteeing the robustness of deep learning models has been a topic of intense research. Ensembling, which combines several classifiers to provide a better model, ha…
Certified private data release for sparse Lipschitz functions
Konstantin Donhauser, Johan Lokna, Amartya Sanyal +3
As machine learning has become more relevant for everyday applications, a natural requirement is the protection of the privacy of the training data. When the relevant learning ques…
Do you pay for Privacy in Online learning?
Amartya Sanyal, Giorgia Ramponi
Online learning, in the mistake bound model, is one of the most fundamental concepts in learning theory. Differential privacy, instead, is the most widely used statistical concept…
A law of adversarial risk, interpolation, and label noise
Daniel Paleka, Amartya Sanyal
In supervised learning, it has been shown that label noise in the data can be interpolated without penalties on test accuracy. We show that interpolating label noise induces advers…
How Robust is Unsupervised Representation Learning to Distribution Shift?
Yuge Shi, Imant Daunhawer, Julia E. Vogt +2
The robustness of machine learning algorithms to distributions shift is primarily discussed in the context of supervised learning (SL). As such, there is a lack of insight on the r…
Catastrophic overfitting can be induced with discriminative non-robust features
Guillermo Ortiz-Jiménez, Pau de Jorge, Amartya Sanyal +5
Adversarial training (AT) is the de facto method for building robust neural networks, but it can be computationally expensive. To mitigate this, fast single-step attacks can be use…
How unfair is private learning ?
Amartya Sanyal, Yaxi Hu, Fanny Yang
As machine learning algorithms are deployed on sensitive data in critical decision making processes, it is becoming increasingly important that they are also private and fair. In t…
Make Some Noise: Reliable and Efficient Single-Step Adversarial Training
Pau de Jorge, Adel Bibi, Riccardo Volpi +4
Recently, Wong et al. showed that adversarial training with single-step FGSM leads to a characteristic failure mode named Catastrophic Overfitting (CO), in which a model becomes su…
Towards Adversarial Evaluations for Inexact Machine Unlearning
Shashwat Goel, Ameya Prabhu, Amartya Sanyal +3
Machine Learning models face increased concerns regarding the storage of personal user data and adverse impacts of corrupted data like backdoors or systematic bias. Machine Unlearn…
Identifying and Exploiting Structures for Reliable Deep Learning
Amartya Sanyal
Deep learning research has recently witnessed an impressively fast-paced progress in a wide range of tasks including computer vision, natural language processing, and reinforcement…