Knowledge-based optimization algorithm for feature selection in promoting circular economy through blockchain.

Blockchain technology support of Circular Economy (CE) concepts offers a promising route to sustainable development in the fields of business, education, healthcare, and the Internet of Things. The integrity of miners is essential to maintaining system reliability in public blockchains. This integrity is threatened by malicious activities like selfish mining and block withholding (BWH). Thus, accurately forecasting honest miners has become a crucial research issue. Although miner behaviour can be modelled by machine learning (ML) classifiers, their effectiveness is heavily reliant on the quality of the input characteristics. Optimal feature selection remains a major challenge in this context. This study addresses this issue using Genetic Algorithms (GA) and Differential Evolution (DE), both proven for optimization tasks. However, conventional GA and DE often suffer from slow convergence and suboptimal solutions. To overcome these limitations, we propose Knowledge-based GA (KBGA) and Knowledge-based DE (KBDE) by embedding domain-specific knowledge. These enhanced methods increase the viability of feature selection solutions and search efficiency. Comprehensive analyses show that KBGA and KBDE predict honest miner behaviour better than conventional GA and DE. The superiority of the suggested methods is further confirmed by statistical studies. This work promotes CE-driven sustainable development endeavours and advances secure Blockchain platforms.