Zero-Knowledge Proof of Training Secures Private Decentralized Machine Learning
ZKPoT consensus uses zk-SNARKs to prove model accuracy privately, resolving the privacy-utility-efficiency trilemma for federated learning.
Optimal Flexible Consensus Allows Client-Specific Safety-Liveness Trade-Offs
This new BFT construction enables clients to optimally select their safety-liveness resilience, fundamentally decentralizing the finality trade-off.
Accountable Byzantine Consensus Achieves Optimal Communication and Accountability Complexity
New Accountable Byzantine Consensus protocol, `abcopt`, delivers optimal communication complexity while guaranteeing provable validator accountability.
Zero-Knowledge Proofs Now Scale Square-Root Memory on Resource-Constrained Devices
A space-efficient tree algorithm cuts ZKP memory from linear to square-root complexity, democratizing verifiable computation on mobile and edge devices.
Inner Product Arguments Eliminate Trusted Setup for Data Availability Sampling
Inner Product Arguments enable trustless data availability sampling by replacing complex trusted setups with a transparent, discrete log-based commitment scheme.
Decentralized Proofs of Encrypted Web Facts without Revealing Underlying Data
DiStefano uses Two-Party Computation within TLS 1.3 to secret-share session keys, enabling zero-knowledge proofs over encrypted web data for private verification.
ZKPoT Consensus Secures Federated Learning with Verifiable, Private Model Contributions
Zero-Knowledge Proof of Training (ZKPoT) is a new consensus primitive that cryptographically verifies model accuracy without exposing private training data, resolving the privacy-utility conflict in decentralized AI.
Validated Strong Consensus Enables Efficient Asynchronous Leader-Based Blockchain State Replication
A new validated strong BFT model allows asynchronous blockchains to use leader-based coordination, achieving HotStuff-level efficiency and linear view changes.
Cryptographic Sortition Achieves Fair Decentralized Transaction Ordering, Mitigating MEV Risk
A new sortition protocol leverages verifiable randomness to select transactions fairly, eliminating the centralized sequencer's ability to extract MEV.
