Results 131 to 140 of about 1,662,189 (295)
Not all adversarial examples require a complex defense : identifying over-optimized adversarial examples with IQR-based logit thresholding [PDF]
, 2019 Detecting adversarial examples currently stands as one of the biggest challenges in the field of deep learning. Adversarial attacks, which produce adversarial examples, increase the prediction likelihood of a target class for a particular data point ...
De Neve, Wesley +5 more
core +1 more source
Ranking the Transferability of Adversarial Examples
ACM Transactions on Intelligent Systems and TechnologyAdversarial transferability in blackbox scenarios presents a unique challenge: while attackers can employ surrogate models to craft adversarial examples, they lack assurance on whether these examples will successfully compromise the target model. Until now, the prevalent method to ascertain success has been trial and error—testing crafted samples ...
Moshe Levy +3 more
openaire +2 more sources
Composition‐Aware Cross‐Sectional Integration for Spatial Transcriptomics
Advanced Intelligent Discovery, EarlyView.Multi‐section spatial transcriptomics demands coherent cell‐type deconvolution, domain detection, and batch correction, yet existing pipelines treat these tasks separately. FUSION unifies them within a composition‐aware latent framework, modeling reads as cell‐type–specific topics and clustering in embedding space.
Qishi Dong +5 more
wiley +1 more source
Adversarial Examples Are Not Bugs, They Are Superposition
CoRRAdversarial examples -- inputs with imperceptible perturbations that fool neural networks -- remain one of deep learning's most perplexing phenomena despite nearly a decade of research. While numerous defenses and explanations have been proposed, there is no consensus on the fundamental mechanism.
Liv Gorton, Owen Lewis
openaire +2 more sources
Harnessing Machine Learning to Understand and Design Disordered Solids
Advanced Intelligent Discovery, EarlyView.This review maps the dynamic evolution of machine learning in disordered solids, from structural representations to generative modeling. It explores how deep learning and model explainability transform property prediction into profound physical insight.
Muchen Wang, Yue Fan
wiley +1 more source
Generating Natural Adversarial Examples
CoRR, 2017 Due to their complex nature, it is hard to characterize the ways in which machine learning models can misbehave or be exploited when deployed. Recent work on adversarial examples, i.e. inputs with minor perturbations that result in substantially different model predictions, is helpful in evaluating the robustness of these models by exposing the ...
Zhengli Zhao +2 more
openaire +4 more sources
Research on Image Adversarial Example Generation Method Based on SE-AdvGAN [PDF]
Jisuanji gongchengAdversarial examples are crucial for evaluating the robustness of Deep Neural Network (DNN) and revealing their potential security risks. The adversarial example generation method based on a Generative Adversarial Network (GAN), AdvGAN, has made ...
ZHAO Hong, SONG Furong, LI Wengai
doaj +1 more source
AI‐Driven Cancer Multi‐Omics: A Review From the Data Pipeline Perspective
Advanced Intelligent Discovery, EarlyView.The exponential growth of cancer multi‐omics data brings opportunities and challenges for precision oncology. This review systematically examines AI's role in addressing these challenges, covering generative models, integration architectures, Explainable AI for clinical trust, clinical applications, and key directions for clinical translation.
Shilong Liu, Shunxiang Li, Kun Qian
wiley +1 more source
Synthesizing Robust Adversarial Examples
CoRR, 2017 Standard methods for generating adversarial examples for neural networks do not consistently fool neural network classifiers in the physical world due to a combination of viewpoint shifts, camera noise, and other natural transformations, limiting their relevance to real-world systems.
Anish Athalye +3 more
openaire +4 more sources
Privacy Protection via Adversarial Examples
, 2022 Machine learning is increasingly exploited by attackers to perform automated, large-scale inference attacks. For instance, in attribute inference attacks, an attacker can use a machine learning classifier to predict a target user's private, sensitive ...
Jia, Jinyuan
core +1 more source