RzkFL: a Verifiable, Fast and Privacy-Preserving Framework for Federated Learning Inference Using Recursive Zero-Knowledge Proofs and on-Chain Verification

Keyphrases

• Federated Learning 100%
• Zero-knowledge Proof 100%
• Privacy-preserving Framework 100%
• Output Layer 66%
• Proof Generation 66%
• Privacy-preserving Machine Learning 66%
• Novae 66%
• Folding Schemes 66%
• Data User 33%
• Global Model 33%
• Private Data 33%
• Input Layer 33%
• Hidden Layer 33%
• Machine Learning Framework 33%
• Resource-constrained 33%
• End-to-end Privacy 33%
• Gas Cost 33%
• Verifiability 33%
• Composable 33%
• File System 33%
• Learning Scenarios 33%
• Trust Assumption 33%
• Cryptographic Proofs 33%
• Incrementally Verifiable Computation 33%
• Succinct Proofs 33%
• Ethereum Smart Contracts 33%
• Verifiable Inference 33%
• Local Inference 33%
• Federated Inference 33%
• Decentralized File Storage 33%
• Decentralized AI Systems 33%
• Sequential Dependencies 33%
• Generation Time 33%
• Pay-per-use 33%
• Verification Cost 33%
• Neural Network Inference 33%

Computer Science

• Privacy Preserving 100%
• Knowledge Proof 100%
• Federated machine learning 100%
• Proof 100%
• Artificial Intelligence 28%
• Privacy Preserving Machine Learning 28%
• Neural Network 14%
• Learning Framework 14%
• Ethereum 14%
• Data Model 14%
• Smart Contract 14%
• Storage System 14%
• Network Inference 14%
• Generation Time 14%
• Inference Computation 14%