[2F4-GS-10h] AI応用：経営と経済

[2F4-GS-10h-04] Prediction of the hashrate in Blockchain using machine learning

〇Iseda Hikoto¹ (1. Niigata university)

Keywords:BlockChain, Hashrate, Machine Laerning

Blockchain is a core that underlies technology cryptocurrencies like such as Bitcoin, and is spreading rapidly in our society as a new secure method to assure the authenticity of information and the redundancy of system. One of the measures to assess of the Blockchain system reliability, which is based on Proof-of-Work, is a "Hashrate". Hashrate represents the necessary hashing calculation per seconds to maintain Blockchain system. It is generally accepted that higher hashrate makes the system more robust and its performance higher. Therefore, the hashrate is used as one of the most important indices to consider implementation of the Blockchain platform into critical social infrastructure such as medical facilities and administrative facilities. This paper presents the results of predicting the hashrate of Bitcoin through univariate and multivariate LSTM (Long-Short term Memory) models. We found the multivariate LSTM outperforms univariate LSTM model for one day ahead prediction, while the univariate model appears to predict the hashrate after a week.