Ruisi Jiang: AI middle office, an accelerator for data analysis and fine computing

Source | Zero One Finance

Author | Zhu Meiyin

Recently, the 2022 Zero One Think Tank Digital Economy Annual Conference with the theme of "Digital Development: Building a New Pattern and Building New Advantages" was held in Shanghai. The summit was hosted by Zero One Finance and Zero One Think Tank, co-sponsored by Lujiazui Magazine and ValueLine Magazine, and provided academic support from the Banking Research Center of central university of finance and economics, the Digital Economy Development Research Group of China Science and Technology System Reform Research Association, the Digital Asset Research Center of The Investment Association of China, the Shanghai Finance and Development Laboratory, and the Hengqin Digital Chain Digital Finance Research Institute.

At the meeting, Dr. Jiang Ruisi, an artificial intelligence scientist at Jiayin Jinke, delivered a keynote speech on "In the ERA OF AI, Technology Helps the Digital Economy - From the Perspective of Marketing Risk Control to the Construction of Fintech AI Middle Office".

Jiayin Jinke artificial intelligence scientist Dr. Ruisi Jiang

Jiang Ruisi first showed a set of data, which shows that the scale of the digital economy has reached 40 trillion yuan in 2020, accounting for about 40% of GDP, and still maintains rapid growth. Jiang Ruisi said that if the digital economy is simply split, it can be divided into digital industrialization and industrial digitalization, digital industrialization is the model of data analysis generated by the business and then empowering the business, while industrial digitalization has several basic necessary conditions: there must be tools for data generation and data generation, data tools for recording and preservation, and the ability to analyze data.

The ability to analyze data is a key step of the current data analysis, Jiang Ruisi used the activity of playing mahjong to make a vivid and interesting analogy to the analysis of data, and said that the production of data value lies in computing, and the evolution process of computing is from the evolution of sensory experience to preliminary statistical analysis to traditional mathematical models, and now it has also entered the process of machine learning, and machine learning models have also become the only way to fine computing.

Then Jiang Ruisi combined with specific application scenarios to explain the advantages of machine learning and AI refined computing, compared with the traditional operation strategy, the marketing strategy established by applying machine learning model can effectively reduce operating costs and improve operational efficiency.

Jiang Ruisi illustrated the efficiency of machine learning and AI through examples, and after using the operating model in a certain scenario, it only used 50% of the previous operation volume but increased the business by 20%. In another specific scenario, if the marketing method using AI+MSG is nearly 4pct more than the marketing method that only uses MSG, and nearly 4.5pct more than any marketing method, the blessing of AI can effectively enhance the efficiency and response rate of marketing. In the anti-fraud scenario, the model is also used to judge its information and generate corresponding strategies.

The development of the model needs to go through the acquisition of data to pre-processing to training and then to the development of the online, which is a repetitive and lengthy process, and there will be a variety of problems, such as slow business response, difficult maintenance and management, lack of monitoring and so on. In view of these pain points, Jiang Ruisi pointed out that the process should be simplified to make it a modular operation, and proposed a solution that can use ai middle platform to deal with these pain points and optimize the corresponding process.

AI Middle Office is a complete set of intelligent model full lifecycle management platform and service configuration system. Based on data platform services, through the sharing and reuse of intelligent services, the management of intelligent service R&D related roles, and the standardization and automation of R&D processes, the rapid construction of personalized intelligent services for front-end business is supported. AI middle office has the ability to maximize the value of the data platform, quickly respond to business needs, provide a unified model development standard format, model training, etc., on the other hand, the AI middle office from the beginning of the number to the model online is completed online, the data does not land, with the characteristics of effective protection of user privacy.

Finally, Jiang Ruisi showed jiayin Jinke's construction of AI middle office, including middle office management, operation monitoring, program design, program model and data processing, etc., and said that Jiayin Jinke has continuously increased its size in the construction of AI middle office and continuously improved its products. In the past ten years since its establishment, Jiayin Jinke has built a cloud service platform and financial risk control system with big data-driven as the core concept, providing systematic solutions for the intelligent transformation of financial services of dozens of small and medium-sized banks, and providing strong digital support for the transformation of traditional banks into open banks.