Machine Learning Security: Threats, Countermeasures, and Evaluations (2020-04-13T00:00:00.000000Z)

TL;DR

This survey systematically analyzes the security issues of machine learning, focusing on existing attacks on machine learning systems, corresponding defenses or secure learning techniques, and security evaluation methods.

Abstract

Machine learning has been pervasively used in a wide range of applications due to its technical breakthroughs in recent years. It has demonstrated significant success in dealing with various complex problems, and shows capabilities close to humans or even beyond humans. However, recent studies show that machine learning models are vulnerable to various attacks, which will compromise the security of the models themselves and the application systems. Moreover, such attacks are stealthy due to the unexplained nature of the deep learning models. In this survey, we systematically analyze the security issues of machine learning, focusing on existing attacks on machine learning systems, corresponding defenses or secure learning techniques, and security evaluation methods. Instead of focusing on one stage or one type of attack, this paper covers all the aspects of machine learning security from the training phase to the test phase. First, the machine learning model in the presence of adversaries is presented, and the reasons why machine learning can be attacked are analyzed. Then, the machine learning security-related issues are classified into five categories: training set poisoning; backdoors in the training set; adversarial example attacks; model theft; recovery of sensitive training data. The threat models, attack approaches, and defense techniques are analyzed systematically. To demonstrate that these threats are real concerns in the physical world, we also reviewed the attacks in real-world conditions. Several suggestions on security evaluations of machine learning systems are also provided. Last, future directions for machine learning security are also presented.

Authors

Mingfu Xue

1 papers

Chengxiang Yuan

1 papers

Heyi Wu

1 papers

TL;DR

Abstract

Authors

References127 items

Advbox: a toolbox to generate adversarial examples that fool neural networks

Februus: Input Purification Defense Against Trojan Attacks on Deep Neural Network Systems

Stateful Detection of Black-Box Adversarial Attacks

Defending Against Neural Network Model Stealing Attacks Using Deceptive Perturbations

Fooling Automated Surveillance Cameras: Adversarial Patches to Attack Person Detection

Neural Cleanse: Identifying and Mitigating Backdoor Attacks in Neural Networks

Online Data Poisoning Attack

Evaluating Differentially Private Machine Learning in Practice

STRIP: a defence against trojan attacks on deep neural networks

A New Backdoor Attack in CNNS by Training Set Corruption Without Label Poisoning

Disguised-Nets: Image Disguising for Privacy-preserving Outsourced Deep Learning

Detecting Backdoor Attacks on Deep Neural Networks by Activation Clustering

Exploring Connections Between Active Learning and Model Extraction

Privacy-Preserving Deep Learning and Inference

Stronger data poisoning attacks break data sanitization defenses

Poisoning Attacks to Graph-Based Recommender Systems

Backdoor Embedding in Convolutional Neural Network Models via Invisible Perturbation

Data Poisoning Attacks against Online Learning

MLCapsule: Guarded Offline Deployment of Machine Learning as a Service

Efficient Deep Learning on Multi-Source Private Data

Model Reconstruction from Model Explanations

How To Backdoor Federated Learning

Towards Data Poisoning Attacks in Crowd Sensing Systems

Evasion-Robust Classification on Binary Domains

Robust Physical-World Attacks on Deep Learning Visual Classification

Fine-Pruning: Defending Against Backdooring Attacks on Deep Neural Networks

PRADA: Protecting Against DNN Model Stealing Attacks

Chronic Poisoning against Machine Learning Based IDSs Using Edge Pattern Detection

Privacy-Preserving Deep Learning via Additively Homomorphic Encryption

ShapeShifter: Robust Physical Adversarial Attack on Faster R-CNN Object Detector

Manipulating Machine Learning: Poisoning Attacks and Countermeasures for Regression Learning

SoK: Security and Privacy in Machine Learning

Scalable Private Learning with PATE

Stealing Hyperparameters in Machine Learning

Turning Your Weakness Into a Strength: Watermarking Deep Neural Networks by Backdooring

A Survey on Security Threats and Defensive Techniques of Machine Learning: A Data Driven View

PoTrojan: powerful neural-level trojan designs in deep learning models

Threat of Adversarial Attacks on Deep Learning in Computer Vision: A Survey

Adversarial Examples: Attacks and Defenses for Deep Learning

Targeted Backdoor Attacks on Deep Learning Systems Using Data Poisoning

Wild Patterns: Ten Years After the Rise of Adversarial Machine Learning

SecureDroid: Enhancing Security of Machine Learning-based Detection against Adversarial Android Malware Attacks

A game-theoretic defense against data poisoning attacks in distributed support vector machines

Improving the Adversarial Robustness and Interpretability of Deep Neural Networks by Regularizing their Input Gradients

Mitigating adversarial effects through randomization

Mitigating Poisoning Attacks on Machine Learning Models: A Data Provenance Based Approach

Robust Linear Regression Against Training Data Poisoning

One Pixel Attack for Fooling Deep Neural Networks

Neural Trojans

Backdoor attacks against learning systems

Adversarial Examples for Malware Detection

Privacy Risk in Machine Learning: Analyzing the Connection to Overfitting

Adversarial Machine Learning in Malware Detection: Arms Race between Evasion Attack and Defense

Towards Poisoning of Deep Learning Algorithms with Back-gradient Optimization

Improving Robustness of ML Classifiers against Realizable Evasion Attacks Using Conserved Features

BadNets: Identifying Vulnerabilities in the Machine Learning Model Supply Chain

Adversarial-Playground: A visualization suite showing how adversarial examples fool deep learning

Model Inversion Attacks for Prediction Systems: Without Knowledge of Non-Sensitive Attributes

Synthesizing Robust Adversarial Examples

NO Need to Worry about Adversarial Examples in Object Detection in Autonomous Vehicles

Towards Deep Learning Models Resistant to Adversarial Attacks

Automated poisoning attacks and defenses in malware detection systems: An adversarial machine learning approach

Certified Defenses for Data Poisoning Attacks

MagNet: A Two-Pronged Defense against Adversarial Examples

DeepSecure: Scalable Provably-Secure Deep Learning

SecureML: A System for Scalable Privacy-Preserving Machine Learning

Adversarial Examples Are Not Easily Detected: Bypassing Ten Detection Methods

Yes, Machine Learning Can Be More Secure! A Case Study on Android Malware Detection

How to steal a machine learning classifier with deep learning

Enhancing robustness of machine learning systems via data transformations

Feature Squeezing: Detecting Adversarial Examples in Deep Neural Networks

Generative Poisoning Attack Method Against Neural Networks

Deep Models Under the GAN: Information Leakage from Collaborative Deep Learning

On Detecting Adversarial Perturbations

Auror: defending against poisoning attacks in collaborative deep learning systems

Practical Secure Aggregation for Federated Learning on User-Held Data