Zobrazeno 1 - 10
of 1 493
pro vyhledávání: '"Kurakin A"'
Publikováno v:
Ukrainian Botanical Journal, Vol 81, Iss 5, Pp 366-373 (2024)
A new record of the rare species Colchicum fominii (Colchicaceae) in Odesa Region is reported. The new locality was recorded on the steppe slope in the Kuchurhan River valley near Hayivka village, Rozdilna District. Its total area was 60,000 m2, with
Externí odkaz:
https://doaj.org/article/cb7c2fc7670b4c649d4cd883674f2774
Autor:
Amin, Kareem, Bie, Alex, Kong, Weiwei, Kurakin, Alexey, Ponomareva, Natalia, Syed, Umar, Terzis, Andreas, Vassilvitskii, Sergei
We present an approach for generating differentially private synthetic text using large language models (LLMs), via private prediction. In the private prediction framework, we only require the output synthetic data to satisfy differential privacy gua
Externí odkaz:
http://arxiv.org/abs/2407.12108
We present a certified defense to clean-label poisoning attacks. These attacks work by injecting a small number of poisoning samples (e.g., 1%) that contain $p$-norm bounded adversarial perturbations into the training data to induce a targeted miscla
Externí odkaz:
http://arxiv.org/abs/2403.11981
Autor:
Wang, Yunjuan, Hazimeh, Hussein, Ponomareva, Natalia, Kurakin, Alexey, Hammoud, Ibrahim, Arora, Raman
Distribution shifts and adversarial examples are two major challenges for deploying machine learning models. While these challenges have been studied individually, their combination is an important topic that remains relatively under-explored. In thi
Externí odkaz:
http://arxiv.org/abs/2402.11120
Differentially private training algorithms like DP-SGD protect sensitive training data by ensuring that trained models do not reveal private information. An alternative approach, which this paper studies, is to use a sensitive dataset to generate syn
Externí odkaz:
http://arxiv.org/abs/2306.01684
Autor:
Carranza, Aldo Gael, Farahani, Rezsa, Ponomareva, Natalia, Kurakin, Alex, Jagielski, Matthew, Nasr, Milad
We address the challenge of ensuring differential privacy (DP) guarantees in training deep retrieval systems. Training these systems often involves the use of contrastive-style losses, which are typically non-per-example decomposable, making them dif
Externí odkaz:
http://arxiv.org/abs/2305.05973
Autor:
Alfimov, Mikhail, Kurakin, Andrey
Publikováno v:
Nucl. Phys. B 998 (2024) 116418
We present the way to continue the bosonic Thirring model or $\beta\gamma$-system with quartic interaction to Minkowski signature, based on the symmetries of this model. It is shown that the considered Minkowski version of the model is one-loop renor
Externí odkaz:
http://arxiv.org/abs/2303.08787
Autor:
Ponomareva, Natalia, Hazimeh, Hussein, Kurakin, Alex, Xu, Zheng, Denison, Carson, McMahan, H. Brendan, Vassilvitskii, Sergei, Chien, Steve, Thakurta, Abhradeep
Publikováno v:
Journal of Artificial Intelligence Research 77 (2023) 1113-1201
ML models are ubiquitous in real world applications and are a constant focus of research. At the same time, the community has started to realize the importance of protecting the privacy of ML training data. Differential Privacy (DP) has become a gold
Externí odkaz:
http://arxiv.org/abs/2303.00654
This paper describes RETVec, an efficient, resilient, and multilingual text vectorizer designed for neural-based text processing. RETVec combines a novel character encoding with an optional small embedding model to embed words into a 256-dimensional
Externí odkaz:
http://arxiv.org/abs/2302.09207