本文在分析了大数据时代在线学习策略的必要性及其面临的挑战的前提下,设计了基于ART神经网络的在线学习分析策略算法,ART神经网络采用以认知和行为模式为基础的无指导的竞争学习算法实现矢量聚类,良好地维持了网络鲁棒性和可调节性的平衡关系,既能够非常灵活的适应新型输入模式产生的变化,同时又可以避免对网络已记忆模式进行篡改。基于ART神经网络模型的在线学习分析策略主要包括识别阶段、比较阶段、学习阶段以及搜索阶段。通过实验设定输入模式,完成在线学习过程进行测试验证。实验结果表明,本文提出的基于ART神经网络的在线学习分析策略算法能够实现对不同输入模式较高正确率的分类,同时可通过警戒参数对网络的分类粒度进行调节,使在线学习策略具有自适应性。 Based on the analysis of the necessity and challenge of online learning strategy in the era of big data, this paper designs an algorithm for online learning and analysis based on ART neural network. ART neural network adopts the guidance of cognitive and behavioral models The competitive learning algorithm implements vector clustering and well maintains the balance between network robustness and adjustability. It not only can adapt to the changes brought by the new input mode very flexibly, but also can avoid tampering with the network memory mode. The online learning and analysis strategy based on ART neural network model mainly includes the recognition phase, the comparison phase, the learning phase and the search phase. Set the input mode through experiments, complete the online learning process to test and verify. The experimental results show that the online learning strategy algorithm based on ART neural network proposed in this paper can classify the higher correct rate of different input modes and adjust the classification granularity of the network through the alert parameters to make the online learning strategy adaptive Sex.