Results 81 to 90 of about 246,254 (290)

secml: Secure and explainable machine learning in Python

SoftwareX, 2022
We present secml, an open-source Python library for secure and explainable machine learning. It implements the most popular attacks against machine learning, including test-time evasion attacks to generate adversarial examples against deep neural ...
Maura Pintor, Luca Demetrio, Angelo Sotgiu, Marco Melis, Ambra Demontis, Battista Biggio   +5 more
doaj   +1 more source

Refinement of amino‐acid conformation vs. difference density maps in time‐resolved serial femtosecond crystallography data analysis

FEBS Open Bio, EarlyView.
The dFoCC pipeline starts with observed DED and resting‐state coordinates, which are then used to generate a library of triggered states. Correlation analysis of the calculated DED features of each candidate vs observed DED permits quantitative evaluation of candidate structural quality.
Meng Iao Fong, Yuhei Hosokawa, Lars Oliver Essen, Manuel Maestre‐Reyna   +3 more
wiley   +1 more source

Generating Hard-Label Black-Box Adversarial Examples for Video Recognition Models

Mathematics
In recent years, video recognition models have witnessed the rapid development of Deep Neural Networks (DNNs). However, these models remain not robust to adversarial examples that are created by adding imperceptible perturbations to clean samples. Recent
Yulin Jing, Lijun Wu, Kaile Su, Wei Wu, Zhiyuan Li, Qi Deng   +5 more
doaj   +1 more source

Adversarial Attacks and Defense in Deep Reinforcement Learning (DRL)-Based Traffic Signal Controllers

IEEE Open Journal of Intelligent Transportation Systems, 2021
Security attacks on intelligent transportation systems (ITS) may result in life-threatening situations. Combining deep neural networks with reinforcement learning (RL) models called DRL shows promising results when applied to urban Traffic Signal Control
Ammar Haydari, Michael Zhang, Chen-Nee Chuah   +2 more
doaj   +1 more source

Molecular characterization of covRS mutations in M1UK Streptococcus pyogenes

FEBS Open Bio, EarlyView.
Group A Streptococcus (GAS) acquires covRS mutations driving a hypervirulent bacterial state, frequently associated with invasive disease‐like necrotizing fasciitis. We demonstrate that the newly emerged M1UK GAS lineage can also acquire these mutations.
Jarrad Pritchard, Emma‐Jayne Proctor, Jiawa Wang, Nikolas P. Johnston, Andrew Hayes, Georgia McCorkell, Haibo Yu, Jai Tree, Mark R. Davies, Mark J. Walker, Ronald Sluyter, Stephan Brouwer, Martina L. Sanderson‐Smith   +12 more
wiley   +1 more source

Opening the AI Black Box: Distilling Machine-Learned Algorithms into Code

Entropy
Can we turn AI black boxes into code? Although this mission sounds extremely challenging, we show that it is not entirely impossible by presenting a proof-of-concept method, MIPS, that can synthesize programs based on the automated mechanistic ...
Eric J. Michaud, Isaac Liao, Vedang Lad, Ziming Liu, Anish Mudide, Chloe Loughridge, Zifan Carl Guo, Tara Rezaei Kheirkhah, Mateja Vukelić, Max Tegmark   +9 more
doaj   +1 more source

Structural and biochemical insights into the thermostable esterase Ta0887 from Thermoplasma acidophilum

FEBS Open Bio, EarlyView.
In this study, a novel esterase from the thermoacidophilic archaeon Thermoplasma acidophilum was biochemically and structurally characterized. Our results demonstrate that Ta0887 is a highly thermostable esterase that preferentially hydrolyzes p‐nitrophenyl hexanoate and possesses an α‐helical cap domain that likely contributes to its substrate ...
Alejandro Delgado‐Rey, Miriam Livier Llamas‐García, Gabriela M. Montero‐Morán, Leticia Santos, Samuel Lara‐González   +4 more
wiley   +1 more source

Efficient black-box attack with surrogate models and multiple universal adversarial perturbations

Scientific Reports
Deep learning models are inherently vulnerable to adversarial examples, particularly in black-box settings where attackers have limited knowledge of the target model.
Tao Ma, Hong Zhao, Ling Tang, Mingsheng Xue, Jing Liu   +4 more
doaj   +1 more source

Real‐World Safety and Effectiveness of JAK Inhibitors in Systemic Sclerosis: A Propensity‐Matched Study From the EUSTAR Cohort

Arthritis Care &Research, EarlyView.
Objective JAK inhibitors (JAKi) have shown promising effects in early‐phase studies of systemic sclerosis (SSc). We aimed to assess the safety and explore the effectiveness of JAKi compared to conventional immunosuppressants in SSc. Methods A longitudinal retrospective study of the European Scleroderma Trials and Research Group (EUSTAR) cohort was ...
Stefano Di Donato, Marie‐Elise Truchetet, Marco Minerba, Oliver Distler, Juan José Alegre Sancho, Yolanda Braun‐Moscovici, Christina Bergmann, Petros Sfikakis, Jeska K. de Vries‐Bouwstra, Murray Baron, Silvia Bellando‐Randone, Lorenzo Dagna, Christopher P. Denton, Madelon C. Vonk, Vanessa Smith, Ivan Castellvi, Gabriela Riemekasten, Andra Balanescu, Masataka Kuwana, Maria De Santis, Kamal Solanki, Anastas Batalov, Vahan Mukuchyan, Marco Matucci‐Cerinic, Yannick Allanore, Francesco Del Galdo, Michael Hughes, EUSTAR Collaborators   +27 more
wiley   +1 more source

dynoGP: Deep Gaussian Processes for Dynamic System Identification

International Journal of Adaptive Control and Signal Processing, EarlyView.
This work introduces a novel class of deep models for system identification, dynamical deep Gaussian processes, which combine the strengths of data‐driven methods, such as those based on neural network architectures, with the ability to output a probability distribution for uncertainty representation.
Alessio Benavoli, Marco Forgione, Dario Piga   +2 more
wiley   +1 more source