A Proof of Useful Work for Artificial Intelligence on the Blockchain

Bitcoin mining is a wasteful and resource-intensive process. To add a block of transactions to the blockchain, miners spend a considerable amount of energy. The Bitcoin protocol, named 'proof of work' (PoW), resembles a lottery and the underlying computational work is not useful otherwise. In this paper, we describe a novel 'proof of useful work' (PoUW) protocol based on training a machine learning model on the blockchain. Miners get a chance to create new coins after performing honest ML training work. Clients submit tasks and pay all training contributors. This is an extra incentive to participate in the network because the system does not rely only on the lottery procedure. Using our consensus protocol, interested parties can order, complete, and verify useful work in a distributed environment. We outline mechanisms to reward useful work and punish malicious actors. We aim to build better AI systems using the security of the blockchain.

[1]  Prashant Nalini Vasudevan,et al.  Proofs of Useful Work , 2017, IACR Cryptol. ePrint Arch..

[2]  Lutz Prechelt,et al.  Early Stopping-But When? , 1996, Neural Networks: Tricks of the Trade.

[3]  Abhinandan Das,et al.  SWIM: scalable weakly-consistent infection-style process group membership protocol , 2002, Proceedings International Conference on Dependable Systems and Networks.

[4]  Miguel Oom Temudo de Castro,et al.  Practical Byzantine fault tolerance , 1999, OSDI '99.

[5]  Zheng Zhang,et al.  MXNet: A Flexible and Efficient Machine Learning Library for Heterogeneous Distributed Systems , 2015, ArXiv.

[6]  Alessandro Duminuco,et al.  Data redundancy and maintenance for peer-to-peer file backup systems , 2009 .

[7]  Jonathan Chiu,et al.  Incentive Compatibility on the Blockchain , 2018, Studies in Economic Design.

[8]  Silvio Micali,et al.  Algorand: Scaling Byzantine Agreements for Cryptocurrencies , 2017, IACR Cryptol. ePrint Arch..

[9]  Michael Backes,et al.  Reliable broadcast in a computational hybrid model with Byzantine faults, crashes, and recoveries , 2003, 2003 International Conference on Dependable Systems and Networks, 2003. Proceedings..

[10]  Silvio Micali,et al.  Verifiable random functions , 1999, 40th Annual Symposium on Foundations of Computer Science (Cat. No.99CB37039).

[11]  Marc'Aurelio Ranzato,et al.  Large Scale Distributed Deep Networks , 2012, NIPS.

[12]  Rachid Guerraoui,et al.  Machine Learning with Adversaries: Byzantine Tolerant Gradient Descent , 2017, NIPS.

[13]  Torben P. Pedersen.  A Threshold Cryptosystem without a Trusted Party (Extended Abstract) , 1991, EUROCRYPT.

[14]  Satoshi Nakamoto.  Bitcoin : A Peer-to-Peer Electronic Cash System , 2009 .

[15]  Geoffrey E. Hinton,et al.  Deep Learning , 2015, Nature.

[16]  Kaiming He,et al.  Accurate, Large Minibatch SGD: Training ImageNet in 1 Hour , 2017, ArXiv.

[17]  Alejandro Baldominos Gómez,et al.  Coin.AI: A Proof-of-Useful-Work Scheme for Blockchain-Based Distributed Deep Learning , 2019, Entropy.

[18]  Robert H. Deng,et al.  CrowdBC: A Blockchain-Based Decentralized Framework for Crowdsourcing , 2019, IEEE Transactions on Parallel and Distributed Systems.

[19]  Elaine Shi,et al.  Permacoin: Repurposing Bitcoin Work for Data Preservation , 2014, 2014 IEEE Symposium on Security and Privacy.

[20]  Hovav Shacham,et al.  Short Signatures from the Weil Pairing , 2001, J. Cryptol..

[21]  Sargur N. Srihari,et al.  Properties of Binary Vector Dissimilarity Measures , 2003 .

[22]  David P. Anderson,et al.  SETI@home: an experiment in public-resource computing , 2002, CACM.

[23]  Markus Schulze,et al.  A new monotonic, clone-independent, reversal symmetric, and condorcet-consistent single-winner election method , 2011, Soc. Choice Welf..

[24]  Nikko Strom,et al.  Scalable distributed DNN training using commodity GPU cloud computing , 2015, INTERSPEECH.

[25]  David R. Karger,et al.  Consistent hashing and random trees: distributed caching protocols for relieving hot spots on the World Wide Web , 1997, STOC '97.

[26]  Morteza Zadimoghaddam,et al.  Consistent Hashing with Bounded Loads , 2016, SODA.

[27]  Tim Roughgarden,et al.  Incentive Compatibility of Bitcoin Mining Pool Reward Functions , 2016, Financial Cryptography.

[28]  F. Moore,et al.  Polynomial Codes Over Certain Finite Fields , 2017 .