Integrating artificial intelligence, operations research, and big data analytics for decision and risk analysis

在信息时代，数据量非常大，并且持续高速增长，为智能决策和风险分析提供了前所未有的机会。然而，这种信息溢出也会引入噪声、精度威胁和性能障碍。由于现有技术效率低下而导致的错误决策可能会导致影响社区的重大成本。此外，大数据的“5V”特征，即卷，多样性，速度，可变性和准确性，需要新的算法来实现增强的决策和知识发现，例如，需要处理实时海量数据。这项研究的目标是整合人工智能，运筹学和大数据分析，以开发高性能的决策支持和风险分析算法。 首先，它旨在解决大数据分析中的关键问题，以找到易于计算，强大和有效的分析方法。我提出了一个基于部分最大相关信息（PMCI）的大数据决策和风险分析框架。其次，本研究将创新性地将DEA与分类器相结合来预测决策单元的相对效率水平。分类器利用样本进行训练，建立样本特征与预先定义的类标签之间的映射关系。经过训练的分类器可以预测新样本的类别标签。第三，人工智能方法将与大数据分析相结合，进行信息融合，以探索业务洞察力和服务经济学。最后，并行计算是处理大数据的有效方法。拟议的研究将实施不同的并行计算框架，以推进使用大数据进行决策的人工智能算法。总体而言，拟议的研究将提供一种创新和独特的方法来解决关键问题，将人工智能，运筹学和大数据分析应用于决策和风险分析。它将为大数据时代的决策和风险分析开辟新的可能性。本研究的目标也是使用图形处理单元（GPU），Hadoop和Apache Spark来并行所提出的算法，以加速其执行。

In the information age, the volume of data is incredibly large and is continually growing at a high rate, offering unprecedented opportunities for intelligent decision making and risk analysis. However, this overflow of information can also introduce noise, accuracy threats and performance obstacles. Wrong decisions due to the inefficiency of existing techniques can cause significant costs affecting the community. Moreover, “5Vs” characteristics of big data, namely Volume, Variety, Velocity, Variability and Veracity, require new algorithms to enable enhanced decision making and knowledge discovery in context where, for example, real-time massive data need to be processed.****The objective of this proposed research is to integrate artificial intelligence, operations research, and big data analytics to develop high performance decision support and risk analysis algorithms. Firstly, it aims at addressing the key issue in big data analytics to find easy-to-calculate, robust and efficient analytical methods. I propose a framework based on partial maximum correlation information (PMCI) for decision and risk analysis in big data. Secondly, the proposed research will creatively integrate DEA and classifiers to predict the relative efficiency level of DMUs. Classifier constructs the mapping relationship between the sample features and the pre-defined class labels after training with samples. The trained classifier can predict the class labels for the new samples. Thirdly, the AI approaches will be combined with big data analytics for information fusion in order to explore business insights and economics of services. Finally, parallel computing is an efficient method to deal with big data. The proposed research will implement different parallel computing frameworks to advance AI algorithms for decision making using big data.****Overall, the proposed research will provide an innovative and unique approach to address key issues to apply artificial intelligence, operations research, and big data analytics for decision and risk analysis. It will open up new possibilities for decision and risk analysis in the big data era. It is also the target of this research to parallel the proposed algorithms using Graphics Processing Unit (GPU), Hadoop and Apache Spark to accelerate their execution.******