GitHub

adversarial-example

Advbox is a toolbox to generate adversarial examples that fool neural networks in PaddlePaddle、PyTorch、Caffe2、MxNet、Keras、TensorFlow and Advbox can benchmark the robustness of machine learning models.
IEEE

TextDefense: Adversarial Text Detection Based on Word Importance Score Dispersion

Abstract: Natural language processing (NLP) models are widely used in various scenarios, yet they are vulnerable to adversarial attacks. Existing works aim to mitigate this vulnerability, but each ...
IEEE

Code Difference Guided Adversarial Example Generation for Deep Code Models

Abstract: Adversarial examples are important to test and enhance the robustness of deep code models. As source code is discrete and has to strictly stick to complex grammar and semantics constraints, ...
note

・敵対的攻撃：訓練後のモデルに摂動を加えたデータ（Adversarial Example;

敵対的サンプル）を入力し、誤分類させること ブラックボックス問題：AIの仕組みや予測の根拠の説明が難しいという問題 ...
Wired

Blind Spots in AI Just Might Help Protect Your Privacy

Machine learning, for all its benevolent potential to detect cancers and create collision-proof self-driving cars, also threatens to upend our notions of what's visible and hidden. It can, for ...
GitHub

JPolasko/Adversarial-Attack-visualization

This repository focuses on visualizing how adversarial attacks (L0, L1, L2, Linf) affect the internal behavior of trained neural networks using methods such as KNN counting and manifold proximity ...
Devdiscourse

Adversarial attacks hit financial AI twice as hard during economic stress

The study, titled Conditional Adversarial Fragility in Financial Machine Learning under Macroeconomic Stress, published as a ...
The Verge

This colorful printed patch makes you pretty much invisible to AI

The patch only fools a specific algorithm, but researchers are working on more flexible solutions The patch only fools a specific algorithm, but researchers are working on more flexible solutions is a ...
ascopubs.org

Using Adversarial Images to Assess the Robustness of Deep Learning Models Trained on ...

HealthTree Cure Hub: A Patient-Derived, Patient-Driven Clinical Cancer Information Platform Used to Overcome Hurdles and Accelerate Research in Multiple Myeloma Adversarial images represent a ...