A Deep Dive into Adversarial Robustness in Zero-Shot Learning (2020-08-17T00:00:00.000000Z)

TL;DR

A study aimed on evaluating the adversarial robustness of ZSL and GZSL models by leveraging the well-established label embedding model and subjecting it to a set of established adversarial attacks and defenses across multiple datasets.

Abstract

Machine learning (ML) systems have introduced significant advances in various fields, due to the introduction of highly complex models. Despite their success, it has been shown multiple times that machine learning models are prone to imperceptible perturbations that can severely degrade their accuracy. So far, existing studies have primarily focused on models where supervision across all classes were available. In constrast, Zero-shot Learning (ZSL) and Generalized Zero-shot Learning (GZSL) tasks inherently lack supervision across all classes. In this paper, we present a study aimed on evaluating the adversarial robustness of ZSL and GZSL models. We leverage the well-established label embedding model and subject it to a set of established adversarial attacks and defenses across multiple datasets. In addition to creating possibly the first benchmark on adversarial robustness of ZSL models, we also present analyses on important points that require attention for better interpretation of ZSL robustness results. We hope these points, along with the benchmark, will help researchers establish a better understanding what challenges lie ahead and help guide their work.

Authors

R. G. Cinbis

7 papers

M. K. Yucel

3 papers

P. Sahin

1 papers

TL;DR

Abstract

Authors

References72 items

A Self-supervised Approach for Adversarial Robustness

One-Shot Adversarial Attacks on Visual Tracking With Dual Attention

Fooling Detection Alone is Not Enough: Adversarial Attack against Multiple Object Tracking

Adversarial Robustness via Label-Smoothing

PyTorch: An Imperative Style, High-Performance Deep Learning Library

Adversarial Attacks and Defenses in Images, Graphs and Text: A Review

Metric Learning for Adversarial Robustness

Transferable Contrastive Network for Generalized Zero-Shot Learning

Adversarial Objects Against LiDAR-Based Autonomous Driving Systems

Defending Against Universal Attacks Through Selective Feature Regeneration

Gradient Matching Generative Networks for Zero-Shot Learning

Unrestricted Adversarial Examples via Semantic Manipulation

Adversarial Attacks on Graph Neural Networks via Meta Learning

Self-Supervised Visual Feature Learning With Deep Neural Networks: A Survey

A Survey of Zero-Shot Learning

Triple Verification Network for Generalized Zero-Shot Learning

Multi-modal Cycle-consistent Generalized Zero-Shot Learning

Adversarial Attacks on Neural Networks for Graph Data

Adversarially Robust Generalization Requires More Data

Defending against Adversarial Images using Basis Functions Transformations

Zero-Shot Recognition via Semantic Embeddings and Knowledge Graphs

Discriminative Learning of Latent Features for Zero-Shot Recognition

Defense-GAN: Protecting Classifiers Against Adversarial Attacks Using Generative Models

Spatially Transformed Adversarial Examples

Audio Adversarial Examples: Targeted Attacks on Speech-to-Text

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

Adversarial Examples: Attacks and Defenses for Deep Learning

HotFlip: White-Box Adversarial Examples for Text Classification

Generalized Zero-Shot Learning via Synthesized Examples

Zero-Shot Visual Recognition Using Semantics-Preserving Adversarial Embedding Networks

On the Robustness of Semantic Segmentation Models to Adversarial Attacks

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

One Pixel Attack for Fooling Deep Neural Networks

Generating Visual Representations for Zero-Shot Classification

Zero-Shot Classification with Discriminative Semantic Representation Learning

Zero-Shot Learning—A Comprehensive Evaluation of the Good, the Bad and the Ugly

Transductive Visual-Semantic Embedding for Zero-shot Learning

MagNet: A Two-Pronged Defense against Adversarial Examples

Black-Box Attacks against RNN based Malware Detection Algorithms

Feature Squeezing: Detecting Adversarial Examples in Deep Neural Networks

SafetyNet: Detecting and Rejecting Adversarial Examples Robustly

Adversarial Examples for Semantic Segmentation and Object Detection

Vulnerability of Deep Reinforcement Learning to Policy Induction Attacks

Perturbations, Optimization, and Statistics

Learning a Deep Embedding Model for Zero-Shot Learning

Delving into Transferable Adversarial Examples and Black-box Attacks

Universal Adversarial Perturbations

Towards Evaluating the Robustness of Neural Networks

Adversarial examples in the physical world

Latent Embeddings for Zero-Shot Classification

Synthesized Classifiers for Zero-Shot Learning

Practical Black-Box Attacks against Machine Learning

Deep Residual Learning for Image Recognition

The Limitations of Deep Learning in Adversarial Settings

Distillation as a Defense to Adversarial Perturbations Against Deep Neural Networks

DeepFool: A Simple and Accurate Method to Fool Deep Neural Networks

Zero-Shot Learning via Semantic Similarity Embedding

Explaining and Harnessing Adversarial Examples

Evaluation of output embeddings for fine-grained image classification

The SUN Attribute Database: Beyond Categories for Deeper Scene Understanding

Intriguing properties of neural networks

DeViSE: A Deep Visual-Semantic Embedding Model

Label-Embedding for Attribute-Based Classification

Zero-Shot Learning Through Cross-Modal Transfer

The Caltech-UCSD Birds-200-2011 Dataset

Large scale image annotation: learning to rank with joint word-image embeddings

Learning to detect unseen object classes by between-class attribute transfer

Countering Adversarial Images using Input Transformations

Attend and Attack : Attention Guided Adversarial Attacks on Visual Question Answering Models

Stacked Semantics-Guided Attention Model for Fine-Grained Zero-Shot Learning

Perturbations , Optimization , and Statistics

Attribute-Based Classification for Zero-Shot Visual Object Categorization

Field of Study