Exploiting Unintended Feature Leakage in Collaborative Learning
Explore this paper's citation graph
Summary
This work shows that an adversarial participant can infer the presence of exact data points -- for example, specific locations -- in others' training data and develops passive and active inference attacks to exploit this leakage.
- Type
- preprint
- Published
- 2018-05-10
- Cited by
- 1,827
- References
- 73
- Access
- Open access
- OpenAlex
- https://openalex.org/W2898291644
- Semantic Scholar
- https://api.semanticscholar.org/CorpusID:53099247
Keywords
Exploit, Computer science, Inference, Adversarial system, Machine learning
References
- NationTelescope: Monitoring and visualizing large-scale collective behavior in LBSNs
- Lasagne: First release.
- Project Adam: Building an Efficient and Scalable Deep Learning Training System
- Privacy in Pharmacogenetics: An End-to-End Case Study of Personalized Warfarin Dosing
- Very Deep Convolutional Networks for Large-Scale Image Recognition
- Labeled Faces in the Wild: A Database forStudying Face Recognition in Unconstrained Environments
- Convolutional Neural Networks for Sentence Classification
- Beyond frontal faces: Improving Person Recognition using multiple cues
- A data-driven approach to cleaning large face datasets
- Resolving Individuals Contributing Trace Amounts of DNA to Highly Complex Mixtures Using High-Density SNP Genotyping Microarrays
- Model Inversion Attacks that Exploit Confidence Information and Basic Countermeasures
- Privacy-preserving deep learning
- Deep learning in neural networks: An overview
- Dropout: a simple way to prevent neural networks from overfitting
- Hacking smart machines with smarter ones: How to extract meaningful data from machine learning classifiers
- On the Difficulties of Disclosure Prevention in Statistical Databases or The Case for Differential Privacy
- Multiparty Differential Privacy via Aggregation of Locally Trained Classifiers
- Petuum: A New Platform for Distributed Machine Learning on Big Data
- Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation
- Parallelized Stochastic Gradient Descent
Cited by
- ML-Leaks: Model and Data Independent Membership Inference Attacks and Defenses on Machine Learning Models
- How To Backdoor Federated Learning
- Privacy-preserving Machine Learning through Data Obfuscation
- LEAF: A Benchmark for Federated Settings
- Comprehensive Privacy Analysis of Deep Learning: Stand-alone and Federated Learning under Passive and Active White-box Inference Attacks
- Federated Machine Learning
- CodedPrivateML: A Fast and Privacy-Preserving Framework for Distributed Machine Learning
- Adversarial Neural Network Inversion via Auxiliary Knowledge Alignment
- Updates-Leak: Data Set Inference and Reconstruction Attacks in Online Learning
- Comprehensive Privacy Analysis of Deep Learning: Passive and Active White-box Inference Attacks against Centralized and Federated Learning
- Differentially Private Learning with Adaptive Clipping
- Reconciling Utility and Membership Privacy via Knowledge Distillation
- Mobile Edge Computing, Blockchain and Reputation-based Crowdsourcing IoT Federated Learning: A Secure, Decentralized and Privacy-preserving System
- Beyond Inferring Class Representatives: User-Level Privacy Leakage From Federated Learning
- A Federated Learning Approach for Mobile Packet Classification
- Federated Learning for Wireless Communications: Motivation, Opportunities, and Challenges
- Oblivious Sampling Algorithms for Private Data Analysis
- GAN-Leaks: A Taxonomy of Membership Inference Attacks against GANs
- A Method of Information Protection for Collaborative Deep Learning under GAN Model Attack
- CrypTFlow: Secure TensorFlow Inference
Related papers
- Membership Inference Attacks Against Machine Learning Models
- SocInf: Membership Inference Attacks on Social Media Health Data With Machine Learning
- How To Backdoor Federated Learning.
- TIPRDC: Task-Independent Privacy-Respecting Data Crowdsourcing Framework for Deep Learning with Anonymized Intermediate Representations
- A co-learning framework for learning user search intents from rule-generated training data
- Privacy-Preserving Personal Model Training
- AttriInfer: Inferring User Attributes in Online Social Networks Using Markov Random Fields
- A Study on Leakage of Private Information in Social Networks and its Preventive Measures
- Adversarial Learning to Compare: Self-Attentive Prospective Customer Recommendation in Location based Social Networks