Search · four archives
Search · four archives
7 papers · ranked by Valyu relevance
Tianyu Liu, Yijia Xiao, Xiao Luo, Hongyu Zhao
Computational methods should be accurate and robust for tasks in biology and medicine, especially when facing different types of attacks, defined as perturbations of benign data that can cause a significant drop in method performance. Therefore, there is a need for robust models that can defend attacks. In this…
Hongjian Zhou, Xinyu Zou, Jinge Wu, Sean Wu + 18 more
Large language models (LLMs) now reach expert-level scores on medical licensing exams, encouraging the assumption that high scores imply safe medical judgment while patients increasingly use them for health advice. We show this assumption is fragile: when misleading context is injected into questions that LLMs…
Daniel Mas Montserrat, Alexander G. Ioannidis
Adversarial attacks can drastically change the output of a method by performing a small change on its input. While they can be a useful framework to analyze worst-case robustness, they can also be used by malicious agents to perform damage in machine learning-based applications. The proliferation of platforms that…
J. Cerri
Managing wildlife populations requires good data. Researchers and policy makers need reliable population estimates and, in case of commercial or recreational harvesting, also trustworthy information about the number of removed individuals. However, auditing schemes are often weak and political or economic pressure…
Steven Golob, Patrick McKeever, Sikha Pentyala, Martine De Cock + 1 more
Single-cell RNA sequencing (scRNA-seq) data is subject to strict access control due to its sensitive nature, motivating the use of synthetic data generation (SDG) for privacy-preserving data sharing. We present the first adversarial privacy attack that performs meaningfully above random guessing against…
Manuel Huth, Jonas Arruda, Roy Gusinow, Lorenzo Contento + 2 more
Federated Learning (FL) is gaining traction in various fields as it enables integrative data analysis without sharing sensitive data, such as in healthcare. However, the risk of data leakage caused by malicious attacks must be considered. In this study, we introduce a novel attack algorithm that relies on being able to…
Kerem Ayoz, Miray Aysen, Erman Ayday, A. Ercument Cicek
Big data era in genomics promises a breakthrough in medicine, but sharing data in a private manner limits the pace of field. Widely accepted “genomic data-sharing beacon” protocol provides a standardized and secure interface for querying the genomic datasets. The data is only shared if the desired information (e.g., a…