Deep Learning Approach for Enhancing Fault Tolerance for Reliable Distributed System
Lokendra Gour · Journal of Emerging Technologies and Innovative Research · 2021
Data in distributed system is dispersed in a structured, unstructured or semi-structured format. Data is distributed across various institutions organizations or individuals. These institutions are analogous to the working nodes. This dispersed data needs to be handled properly in two respects one for ensuring data security and privacy another is enhancing fault tolerance. Data security and privacy has become a prime concern for data centric applications. Robust fault tolerant platform is required for smooth functioning of the distributed system. The intuitive and robust model FedLearning is framed for distributed learning. FedLearning is based on ensemble learning, in which neural network models are deployed independently on each data unit at the local working node. All the local working model’s parameters are combined and collected by the secure coordinating nodes. Federated learning performs the coordination among the working nodes. Coordination prevents the failures of individual working machines.