SYSTEM FOR DECENTRALIZED AND DISTRIBUTED DEEP LEARNING

摘要

The embodiments herein disclose a system (100) comprising a requester node (101) and a master node (102). The requester node (101) interacts directly with a user or through a computing device. The master node (102) integrates a data received from and transmitted to a plurality of sources. The master node (101) comprises a request pooler (103), a data collection node (104), a data repository (107), and a deep learning training distribution (DLTD) node (109). The request pooler (103) is connected to the requester node (101) through a unidirectional channel. The data collection node (104) is connected to a data source node (105) and a data annotator (106). The data repository (107) comprises a plurality of data sets collected from the data collection node (108). The DLTD node (109) is connected to plurality of computer nodes (110) and the data repository (107).