APPARATUS AND METHOD OF ARITICIAL INTELLIGENCE PREDICTIVE MODEL BASED ON DIPERSION PARALLEL

摘要

Distributed parallel-based artificial intelligence prediction model control apparatus and method of the present invention, in the distributed parallel-based artificial intelligence prediction model control apparatus applied to an artificial intelligence platform and communicating with a learning server cluster, receiving a learning task, the learning task a learning execution scheduling unit for extracting learning attribute information from and estimating the processing workload of the learning task based on the extracted learning attribute information; a parallel execution management unit for performing parallelization processing on the plurality of learning tasks when the learning execution scheduling unit receives a plurality of learning tasks; a distributed execution management unit for determining the learning server to which the learning task is allocated in consideration of the predicted amount of processing workload of the learning task; a learning work resource management unit for calculating an actual processing workload of the learning task allocated to the learning server and generating billing information based on the actual processing workload; and a cluster management unit that collects optimization types and idle capacities of a plurality of learning servers from the learning server cluster and generates server state information.