Incremental Learning Framework Using Cloud Computing

High volume of data, perceived as either challenge or opportunity. Deep learning architecture demands high volume of data to effectively back propagate and train the weights without bias. At the same time, large volume of data demands higher capacity of the machine where it could be executed seamlessly. Budding data scientist along with many research professionals face frequent disconnection issue with cloud computing framework (working without dedicated connection) due to free subscription to the platform. Similar issues also visible while working on local computer where computer may run out of resource or power sometimes and researcher has to start training the models all over again. In this paper, we intend to provide a way to resolve this issue and progressively training the neural network even after having frequent disconnection or resource outage without loosing much of the progress