SOTAVerified

Model extraction

Model extraction attacks, aka model stealing attacks, are used to extract the parameters from the target model. Ideally, the adversary will be able to steal and replicate a model that will have a very similar performance to the target model.

Papers

Recently AddedMost HypedMost ActiveNeeds VerificationMost Verified

Showing 1–50 of 176 papers

TitleStatusHype
Safety at Scale: A Comprehensive Survey of Large Model SafetyCode3
Are You Copying My Model? Protecting the Copyright of Large Language Models for EaaS via Backdoor WatermarkCode1
ATOM: A Framework of Detecting Query-Based Model Extraction Attacks for Graph Neural NetworksCode1
MEA-Defender: A Robust Watermark against Model Extraction AttackCode1
Protecting Language Generation Models via Invisible WatermarkingCode1
Neural Honeytrace: A Robust Plug-and-Play Watermarking Framework against Model Extraction AttacksCode1
MEME: Generating RNN Model Explanations via Model ExtractionCode1
Cryptanalytic Extraction of Neural Network ModelsCode1
Model Extraction and Adversarial Transferability, Your BERT is Vulnerable!Code1
Entangled Watermarks as a Defense against Model ExtractionCode1
"Yes, My LoRD." Guiding Language Model Extraction with Locality Reinforced DistillationCode1
Black-Box Attacks on Sequential Recommenders via Data-Free Model ExtractionCode1
FedRolex: Model-Heterogeneous Federated Learning with Rolling Sub-Model ExtractionCode1
MARLeME: A Multi-Agent Reinforcement Learning Model Extraction LibraryCode1
Watermarking Vision-Language Pre-trained Models for Multi-modal Embedding as a ServiceCode1
Now You See Me (CME): Concept-based Model ExtractionCode1
MEME: Generating RNN Model Explanations via Model ExtractionCode1
Data-Free Model ExtractionCode1
Stateful Detection of Model Extraction AttacksCode0
Safe and Robust Watermark Injection with a Single OoD ImageCode0
Stealing Machine Learning Models via Prediction APIsCode0
Protecting Intellectual Property of Language Generation APIs with Lexical WatermarkCode0
On the Difficulty of Defending Self-Supervised Learning against Model ExtractionCode0
Defense Against Model Extraction Attacks on Recommender SystemsCode0
VidModEx: Interpretable and Efficient Black Box Model Extraction for High-Dimensional SpacesCode0
Process Extraction from Text: Benchmarking the State of the Art and Paving the Way for Future ChallengesCode0
On the Effectiveness of Dataset Watermarking in Adversarial SettingsCode0
Stealing and Evading Malware Classifiers and Antivirus at Low False Positive ConditionsCode0
SAME: Sample Reconstruction against Model Extraction AttacksCode0
Beyond Slow Signs in High-fidelity Model ExtractionCode0
Model extraction from counterfactual explanationsCode0
MISLEADER: Defending against Model Extraction with Ensembles of Distilled ModelsCode0
Model Reconstruction Using Counterfactual Explanations: A Perspective From Polytope TheoryCode0
ACTIVETHIEF: Model Extraction Using Active Learning and Unannotated Public DataCode0
Army of Thieves: Enhancing Black-Box Model Extraction via Ensemble based sample selectionCode0
MeaeQ: Mount Model Extraction Attacks with Efficient QueriesCode0
Knowledge Distillation-Based Model Extraction Attack using GAN-based Private Counterfactual ExplanationsCode0
CEGA: A Cost-Effective Approach for Graph-Based Model Extraction and AcquisitionCode0
Marich: A Query-efficient Distributionally Equivalent Model Extraction Attack using Public DataCode0
From Counterfactuals to Trees: Competitive Analysis of Model Extraction AttacksCode0
DAWN: Dynamic Adversarial Watermarking of Neural NetworksCode0
GUIDO: A Hybrid Approach to Guideline Discovery & Ordering from Natural Language TextsCode0
Robust and Minimally Invasive Watermarking for EaaSCode0
Deep Neural Network Fingerprinting by Conferrable Adversarial ExamplesCode0
An Approach for Process Model Extraction By Multi-Grained Text ClassificationCode0
Model Extraction Attacks on Graph Neural Networks: Taxonomy and RealizationCode0
Efficient and Effective Model ExtractionCode0
FLuID: Mitigating Stragglers in Federated Learning using Invariant DropoutCode0
A Hard-Label Cryptanalytic Extraction of Non-Fully Connected Deep Neural Networks using Side-Channel AttacksCode0
Not Just Change the Labels, Learn the Features: Watermarking Deep Neural Networks with Multi-View DataCode0
Show:102550
← PrevPage 1 of 4Next →

Benchmark Results

#ModelMetricClaimedVerifiedStatus
1three-step-originalExact Match0.17—Unverified