Revolutinoning AI: Artificial Intelligence Based Crypto Currency Farm Mining Application Design Using Federated Deep Learning Principles
G. Ramkumar, S. Jency
Abstract
In this research, an artificial intelligence cryptocurrency mining application using federated deep learning introduced. Federated Learning Framework for Edge Computing The solution mitigates the issue of decentralized data processing, privacy preserving, and latency reduction via a practical federated learning framework. The core of the system is a mixed deep learning model utilizing ResNet50 and SqueezeNet architectures, jointly rendering high accuracy with low computational cost. Taking the federated learning framework as an example, a central server and multiple client nodes are set up, dividing secure communication protocols to ensure integrity of data. Several custom federated learning algorithms were implemented to adjust the decentralized nature of crypto-jacking for local training on client nodes and central server diffusion of model updates. The hybrid model provides a high level of performance in terms of optimizing mining operations with an accuracy rate of $\mathbf{9 6. 2 5 \%}$. This model stored countless data such as hash rates, power consumption and network data that mining activity produced, thus making it possible for the mining strategy to be constantly updated. By integrating the model with mine application various interfaces was developed with user-friendly features, thereby facilitating ease in monitoring and control of mining operations.