Research On Data Processing Model Based On Machine Learning In Cognitive Computing Systems

With the explosive increment of Internet data, it is increasingly difficult to obtain valuable information from huge amounts of this data. As an important branch of Artificial Intelligence(AI), cognitive computing technology plays a key role in Internet applications. Besides, how to realize the efficient data processing is one of the most important tasks of cognitive computing. Thus, this thesis is mainly about the data processing model based on machine learning in cognitive computing systems.Firstly, a cognitive computing model is presented and data processing technology and task scheduling strategy are studied. To sum up, the main contents and innovations of this thesis are included in the following several respects:Due to the difficulty of traditional data processing modes cannot meet the demand of real-time perceptive information services, a cognitive computing model is proposed in this thesis. The cognitive computing model can realize functions of data processing. Moreover, to improve the classification accuracy of Very Fast Decision Tree algorithm, a new decision tree algorithm is presented.Focusing on the judgment of the effectiveness of information, a decision algorithm based on deep belief networks and linear perceptron is proposed. The proposed algorithm has error control capabilities.Due to the classification of resources are not mentioned in the RAM(Reward optim Al Matching) algorithm, which can have negative performance implications. To avoid this, an improved task scheduling algorithm based on matching with queues is presented. The algorithm is finished by two steps. The first step is make clustering analysis for computing resources and the second step is task scheduling.