ZKP Frontier Tracker 🎯
零知识证明 前沿热点追踪 🎯
Thu 星期四
12.11
2025
https://arxiv.org/abs/2510.06784
Zakharov et al. proposed a zkML framework Bionetta based on UltraGroth, significantly improving the proof efficiency of custom neural networks and supporting client-side proofs on mobile devices. Zakharov等人在论文中提出了一种基于UltraGroth的zkML框架Bionetta，显著提升了自定义神经网络的证明效率，支持在移动设备上运行客户端证明。
Notes
- Bionetta based on UltraGroth protocol, optimizing zkML proof performance
- Supports custom neural networks on mobile devices for client-side proofs
- Significantly improves proof time compared to EZKL, Lagrange's deep-prove, etc.
- Deployable on native EVM smart contracts, without large proof scale and verification overhead
- One-time preprocessing steps (circuit compilation, trusted setup) cost increase
- Suitable for client-side machine learning applications requiring privacy protection