Stable Neural ODE with Lyapunov-Stable Equilibrium Points for Defending Against Adversarial Attacks

Last update: Dec 12, 2022

Related tags

Deep Learning SODEF

Overview

Stable Neural ODE with Lyapunov-Stable Equilibrium Points for Defending Against Adversarial Attacks

Stable Neural ODE with Lyapunov-Stable Equilibrium Points for Defending Against Adversarial Attackss.

Qiyu Kang, Yang Song, Qinxu Ding, Wee Peng Tay

Environment settings

OS: Ubuntu 18.04
GPU: RTX 2080 Ti, RTX a5000, RTX 3090
Cuda: 11.1 or 10.2
Python: >=3.6
PyTorch: >= 1.6.0
Torchvision: >= 0.7.0

Empirical Evaluations

Compatibility of SODEF

In this section, we show compatibility of SODEF using TRADES:

We append our SODEF after TRADES net to improve the model robustness against adversarial attacks. TRADES works as the feature extractor $h_{\boldsymbol{\phi}}$ as in our paper. Please note TRADES weights are kept fixed during the training. We use the pretrained model provided by TRADES Repo.

Attack / Model	TRADES ℒ_∞	TRADES+SODEF ℒ_∞	TRADES ℒ₂	TRADES+SODEF ℒ₂
Clean	85.48	85.18	85.48	85.18
APGD_CE	56.08	70.90	61.74	74.35
APGD_DLR^T	53.70	64.15	59.22	68.55
FAB^T	54.18	82.92	60.31	83.15
Square	59.12	62.21	72.65	76.02
AutoAttack	53.69	57.76	59.42	67.75

Tab 1. Classification accuracy (%) using TRADES (w/ and w/o SODEF) under AutoAttack on adversarial CIFAR10 examples with ℒ₂ norm (ϵ = 0.5) and ℒ_∞ norm (ϵ = 8/255).

Transfer attack:

Classification accuracy for adv examples generated from original pretrained model using AA ℒ_∞ (ϵ = 8/255) attacks : 61.94%.

cd trades_r
python sodef_eval_ode.py

cd trades_r
sodef_eval_transfer.ipynb

Notification

More test code and models will be uploaded soon after packing.

We currenly only upload the test code for SODEF. Please understand we have strict protocols for code release as this research is partially funded by corporate funding. We will upload the training code as soon as permission is granted.

Stable Neural ODE with Lyapunov-Stable Equilibrium Points for Defending Against Adversarial Attacks

Related tags

Overview

Stable Neural ODE with Lyapunov-Stable Equilibrium Points for Defending Against Adversarial Attacks

Environment settings

Empirical Evaluations

Compatibility of SODEF

In this section, we show compatibility of SODEF using TRADES:

Notification

Owner

Kang Qiyu

Loopy belief propagation for factor graphs on discrete variables, in JAX!

StarGAN - Official PyTorch Implementation (CVPR 2018)

Relative Uncertainty Learning for Facial Expression Recognition

TiP-Adapter: Training-free CLIP-Adapter for Better Vision-Language Modeling

A Bayesian cognition approach for belief updating of correlation judgement through uncertainty visualizations

An implementation of a discriminant function over a normal distribution to help classify datasets.

Punctuation Restoration using Transformer Models for High-and Low-Resource Languages

OpenDILab RL Kubernetes Custom Resource and Operator Lib

The official repo of the CVPR 2021 paper Group Collaborative Learning for Co-Salient Object Detection .

Creating multimodal multitask models

Code for 2021 NeurIPS --- Towards Multi-Grained Explainability for Graph Neural Networks

Implementation of "StrengthNet: Deep Learning-based Emotion Strength Assessment for Emotional Speech Synthesis"

7th place solution of Human Protein Atlas - Single Cell Classification on Kaggle

Group R-CNN for Point-based Weakly Semi-supervised Object Detection (CVPR2022)

Pytorch Implementation of Zero-Shot Image-to-Text Generation for Visual-Semantic Arithmetic

Contextual Attention Localization for Offline Handwritten Text Recognition

なりすまし検出(anti-spoof-mn3)のWebカメラ向けデモ

The self-supervised goal reaching benchmark introduced in Discovering and Achieving Goals via World Models

OptaPlanner wrappers for Python. Currently significantly slower than OptaPlanner in Java or Kotlin.

For encoding a text longer than 512 tokens, for example 800. Set max_pos to 800 during both preprocessing and training.