Paper proposes Proof-of-ML-Inference to replace hash mining
Original title: Proof-of-Machine-Learning-Inference: Mining Knowledge, Not Hashes
Miners would run requested model inference; a simulation kept block times near the calibrated target.
A cs.CR paper presents Proof-of-ML-Inference (PoML), a consensus protocol in which miners perform user-requested machine-learning inference instead of Bitcoin-style hash puzzles. The authors formalize the assumptions needed for inference to replace hash work and say LLMs and diffusion models satisfy them. They reduce protocol security to the Bitcoin Backbone model and report that a simulation kept block production close to its calibrated target.
Unconfirmed: Evidence that LLMs and diffusion models meet the stated assumptions is the authors’ claim, not an independent confirmation.
- PoML is meant to secure a permissionless chain while serving inference demand.
- Security is argued by reduction to the Bitcoin Backbone model.
- The block-time result is from a simulated environment, not a live network.
From the original article
1 min read at arxiv.orgBitcoin-style Proof-of-Work (PoW) secures permissionless blockchains but spends substantial computation on hash puzzles which have no utility other than extending the chain. At the same time, we have seen that the rise of generative AI has created significant demand for computational resources that are unfortunately available only to a few players. We propose a solution that solves both problems at the same time.
By Wan Ki Wong · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.