Demonstrates iterative FGSM on Apple's NeuralHash model.

Overview

apple-neuralhash-attack

Demonstrates iterative FGSM on Apple's NeuralHash model.

TL;DR: It is possible to apply noise to CSAM images and make them look like regular images to the NeuralHash model. The noise does degrade the CSAM image (see samples). But this was achieved without tuning learning rate and there are more refined attacks available too.

Example

Here is an example that uses a Grumpy Cat image in place of a CSAM image. The attack adds noise to the Grumpy Cat image and makes the model see it as a Doge image.

As a result, both of these images have the same neural hash of 11d9b097ac960bd2c6c131fa, computed via ONNX Runtime, with the script by AsuharietYgvar/AppleNeuralHash2ONNX.

doge adv_cat

More generally, because the attack optimizes the model output, the adversarial image will generate largely the same hash as the good image, regardless of the seed.

Instructions

Get ONNX model

Obtain the ONNX model from AsuharietYgvar/AppleNeuralHash2ONNX. You should have a path to a model.onnx file.

Convert ONNX model to TF model

Then convert the ONNX model to a Tensorflow model by first installing the onnx_tf library via onnx/onnx-tensorflow. Then run the following:

python3 convert.py -o /path/to/model.onnx

This will save a Tensorflow model to the current directory as model.pb.

Run adversarial attack

Finally, run the adversarial attack with the following:

python3 nnhash_attack.py --seed /path/to/neuralhash_128x96_seed1.dat

Other arguments:

-m           Path to Tensorflow model (defaults to "model.pb")
--good       Path to good image (defaults to "samples/doge.png")
--bad        Path to bad image (defaults to "samples/grumpy_cat.png")
--lr         Learning rate (defaults to 3e-1)
--save_every Save every interval (defaults to 2000)

This will save generated images to samples/iteration_{i}.png.

Note that the hash similarity may decrease initially before increasing again.

Also, for the sample images and with default parameters, the hash was identical after 28000 iterations.

Terminal output:

# Some Tensorflow boilerplate...
Iteration #2000: L2-loss=134688, Hash Similarity=0.2916666666666667
Good Hash: 11d9b097ac960bd2c6c131fa
Bad Hash : 20f1089728150af2ca2de49a
Saving image to samples/iteration2000.png...
Iteration #4000: L2-loss=32605, Hash Similarity=0.41666666666666677
Good Hash: 11d9b097ac960bd2c6c131fa
Bad Hash : 20d9b097ac170ad2cfe170da
Saving image to samples/iteration4000.png...
Iteration #6000: L2-loss=18547, Hash Similarity=0.4166666666666667
Good Hash: 11d9b097ac960bd2c6c131fa
Bad Hash : 20d9b097ac170ad2c7c1f0de
Saving image to samples/iteration6000.png...

Credit

Owner
Lim Swee Kiat
Lim Swee Kiat
duralava is a neural network which can simulate a lava lamp in an infinite loop.

duralava duralava is a neural network which can simulate a lava lamp in an infinite loop. Example This is not a real lava lamp but a "fake" one genera

Maximilian Bachl 87 Dec 20, 2022
The spiritual successor to knockknock for PyTorch Lightning, get notified when your training ends

Who's there? The spiritual successor to knockknock for PyTorch Lightning, to get a notification when your training is complete or when it crashes duri

twsl 70 Oct 06, 2022
Identify the emotion of multiple speakers in an Audio Segment

MevonAI - Speech Emotion Recognition Identify the emotion of multiple speakers in a Audio Segment Report Bug · Request Feature Try the Demo Here Table

Suyash More 110 Dec 03, 2022
Normalizing Flows with a resampled base distribution

Resampling Base Distributions of Normalizing Flows Normalizing flows are a popular class of models for approximating probability distributions. Howeve

Vincent Stimper 24 Nov 03, 2022
A coin flip game in which you can put the amount of money below or equal to 1000 and then choose heads or tail

COIN_FLIPPY ##This is a simple example package. You can use Github-flavored Markdown to write your content. Coinflippy A coin flip game in which you c

2 Dec 26, 2021
Hand Gesture Volume Control is AIML based project which uses image processing to control the volume of your Computer.

Hand Gesture Volume Control Modules There are basically three modules Handtracking Program Handtracking Module Volume Control Program Handtracking Pro

VITTAL 1 Jan 12, 2022
Class-Balanced Loss Based on Effective Number of Samples. CVPR 2019

Class-Balanced Loss Based on Effective Number of Samples Tensorflow code for the paper: Class-Balanced Loss Based on Effective Number of Samples Yin C

Yin Cui 546 Jan 08, 2023
ZSL-KG is a general-purpose zero-shot learning framework with a novel transformer graph convolutional network (TrGCN) to learn class representation from common sense knowledge graphs.

ZSL-KG is a general-purpose zero-shot learning framework with a novel transformer graph convolutional network (TrGCN) to learn class representa

Bats Research 94 Nov 21, 2022
Gated-Shape CNN for Semantic Segmentation (ICCV 2019)

GSCNN This is the official code for: Gated-SCNN: Gated Shape CNNs for Semantic Segmentation Towaki Takikawa, David Acuna, Varun Jampani, Sanja Fidler

859 Dec 26, 2022
nextPARS, a novel Illumina-based implementation of in-vitro parallel probing of RNA structures.

nextPARS, a novel Illumina-based implementation of in-vitro parallel probing of RNA structures. Here you will find the scripts necessary to produce th

Jesse Willis 0 Jan 20, 2022
With this package, you can generate mixed-integer linear programming (MIP) models of trained artificial neural networks (ANNs) using the rectified linear unit (ReLU) activation function

With this package, you can generate mixed-integer linear programming (MIP) models of trained artificial neural networks (ANNs) using the rectified linear unit (ReLU) activation function. At the momen

ChemEngAI 40 Dec 27, 2022
Code for "On the Effects of Batch and Weight Normalization in Generative Adversarial Networks"

Note: this repo has been discontinued, please check code for newer version of the paper here Weight Normalized GAN Code for the paper "On the Effects

Sitao Xiang 182 Sep 06, 2021
Implementing SYNTHESIZER: Rethinking Self-Attention in Transformer Models using Pytorch

Implementing SYNTHESIZER: Rethinking Self-Attention in Transformer Models using Pytorch Reference Paper URL Author: Yi Tay, Dara Bahri, Donald Metzler

Myeongjun Kim 66 Nov 30, 2022
Elastic weight consolidation technique for incremental learning.

Overcoming-Catastrophic-forgetting-in-Neural-Networks Elastic weight consolidation technique for incremental learning. About Use this API if you dont

Shivam Saboo 89 Dec 22, 2022
2020 CCF大数据与计算智能大赛-非结构化商业文本信息中隐私信息识别-第7名方案

2020CCF-NER 2020 CCF大数据与计算智能大赛-非结构化商业文本信息中隐私信息识别-第7名方案 bert base + flat + crf + fgm + swa + pu learning策略 + clue数据集 = test1单模0.906 词向量

67 Oct 19, 2022
A working implementation of the Categorical DQN (Distributional RL).

Categorical DQN. Implementation of the Categorical DQN as described in A distributional Perspective on Reinforcement Learning. Thanks to @tudor-berari

Florin Gogianu 98 Sep 20, 2022
Topic Modelling for Humans

gensim – Topic Modelling in Python Gensim is a Python library for topic modelling, document indexing and similarity retrieval with large corpora. Targ

RARE Technologies 13.8k Jan 03, 2023
A flexible and extensible framework for gait recognition.

A flexible and extensible framework for gait recognition. You can focus on designing your own models and comparing with state-of-the-arts easily with the help of OpenGait.

Shiqi Yu 335 Dec 22, 2022
SEJE Pytorch implementation

SEJE is a prototype for the paper Learning Text-Image Joint Embedding for Efficient Cross-Modal Retrieval with Deep Feature Engineering. Contents Inst

0 Oct 21, 2021
Based on the given clinical dataset, Predict whether the patient having Heart Disease or Not having Heart Disease

Heart_Disease_Classification Based on the given clinical dataset, Predict whether the patient having Heart Disease or Not having Heart Disease Dataset

Ashish 1 Jan 30, 2022