基于神经网络算法的大数据分析方法研究

Study on the analysis method of data based on neural network algorithm

随着深度互联网时代到来,大数据所蕴含的巨大科学、经济价值逐渐凸显。然而其数据分析方法却存在较高技术壁垒,想要发掘出大数据的价值空间,需要摒弃传统方案,采用新的分析方法。深度神经网络算法采用仿生学习算法整合庞大的异构数据,支持多源信息筛选,可实现时序动态捕捉,从而搭建起大数据转化为价值信息的桥梁。文中着重分析"大数据+神经网络"的深度学习算法在非结构化、模式多变的大数据群中的特征提取模式;并基于无限神经网络的前馈式连接方法,耦合时间参数进行更精确的特征提取与数据预测。最后对其在语音识别和图像分析中的应用进行实例测试,数据结果表明:无限神经网络在数据处理中具备更为强大的计算效率和性能优势。

With the advent of the depth of the Internet era,the great economic and scientific value of the big data is gradually highlighted. However,there are technical barriers to the data analysis methods of big data. In order to explore the value space of big data,we need to abandon the traditional program and develop new data analysis methods. The deep neural network algorithm uses bionic learning algorithm to integrate huge heterogeneous data,filters multi- source information,and realizes dynamic capture,which can perfect the bridge of transforming big data into value information. This paper focuses on the analysis of ＂big data ＋ neural network＂ deep learning algorithm in unstructured model,changeable,characteristics of cross domain data in extraction mode,and feedforward neural network based on infinite connection method,coupling time parameter prediction feature extraction and more accurate data. The final test of its application in speech recognition and image analysis,the results show that the infinite neural network in data processing compared with the ordinary algorithm have more computational efficiency and powerful performance advantage.