Projeto Oficial do Módulo de Controle de Desempenho e Utilidades do SEI. PROJETO DE DESENVOLVIMENTO COLABORATIVO, contudo, deve abrir Issue de demanda para discussão antes de qualquer desenvolvimento.

Projected Gradient Descent (PGD), inverted and amplified -> prompt & generate images with CLIP

PyTorch implementation of projected gradient descent (PGD) adversarial noise attack

A resources of PGD class tasks. With this repository, you can practice pgdit class and home tasks run them based on your requirements.

Defence of Five

some tricks for sentence representation

PGD GameDev JAM #10: Revived Fears

Programa de Gestão e Desempenho (PGD) versão SESEP (Secretaria de Avaliação e Gestão da Informação) em Docker/container.

Code implementation of our AAAI'22 paper "Improved Gradient-Based Adversarial Attacks for Quantized Networks"

对抗测试

Homework of Security and Privacy of Machine Learning (SPML Lectured by Shang-Tse Chen at NTU)

Official implementation of the paper: "REGroup: Rank-aggregating Ensemble of Generative Classifiers for Robust Predictions", IEEE WACV, 2022

Paddle-Adversarial-Toolbox (PAT) is a Python library for Deep Learning Security based on PaddlePaddle.

Implementation of gradient-based adversarial attack(FGSM,MI-FGSM,PGD)

Gradient Decent, Projected Gradient Decent and Stochastic Gradient Decent implementaion in python

Pytorch-Named-Entity-Recognition-with-transformers

Implementation of adversarial training under fast-gradient sign method (FGSM), projected gradient descent (PGD) and CW using Wide-ResNet-28-10 on cifar-10. Sample code is re-usable despite changing the model or dataset.

Reproduce multiple adversarial attack methods

Package for adversarial attack in pytorch

Structural Wing Model Reduction in Fluid-Structure Interactions