close
AI

冰鉴科技战略合作部总经理管琯:AI落地金融,引领智能风控 | 2019 WISE 超级进化者大会

bihog9ib1y4ihxgr.jpeg

On July 9-10, 36氪 held the “2019WISE Super Evolutionist” conference in Beijing and Shanghai. The event has seven conference venues, focusing on the path of enterprise development and change, the grasp of industry trends, the advancement and transformation of the retail industry, and the trillions of enterprises. Topics such as the rise of the service market, industrial innovation opportunities, globalization trends and the outbreak of differentiated demand invite more than 100 industry leaders to focus on the rise of super-evolutionists who lead the industry. In the era of mobile Internet, with the development of Internet finance, the long-tailed consumers who occupy about 60%-80% of China’s population, and the financial needs of small and micro enterprises have become increasingly prominent. When financial institutions serve these long-tailed groups, they generally face two problems. One is the lack of data analysis talents in core risk control, and the other is that online services are at a disadvantage. As a third-party organization, Bingjian Technology hopes to use AI technology to help inclusive finance. Guan Yi, general manager of the ice-tech technology strategic cooperation department and general manager of the national new financial sales department, believes that the financial needs of the long-tailed people are characterized by small amount and high frequency, which is good for data and algorithms, only in such a scenario. Only when there are enough samples to train can you use the model to realize its value. Through the cooperation with financial institutions, Bingjian Technology can obtain the target variables needed for a large number of samples. At the same time, it is also possible to integrate mobile Internet scenarios to help small and micro enterprises and individuals conduct credit evaluation. This is also due to the rich accumulation of data from non-financial institutions, individuals and small and micro enterprises in our country. Whether it is e-commerce, consumption, communication or geographical location, many of them are concentrated in the mobile Internet and mobile phones. These data will be well applied in the future in the context of inclusive finance. The following is a guest speech: I heard a lot of guests sharing AI, big data, and the concept that everyone has been listening for a long time. Behind these concepts, there is not much difference in the technical principle itself. The difference between the two is more about how to apply these technologies and combine business to create different social values. Some of the guests just shared the application of AI and models in various fields, including logistics, speech recognition, recommendation algorithms, and IOT. The theme I shared today is how Ice Technology uses AI technology to help inclusive finance. The story I told, in addition to AI technology, one is the mobile Internet, and the other is inclusive finance. Ice Jian Technology is an independent third-party risk control service provider. What we do is to help financial institutions, banks, consumer finance companies, and small loan companies conduct credit evaluations of individuals and small and micro enterprises. These organizations decide whether they can give individuals according to the assessment. Credit loans are provided to small and micro enterprises. Why is there a company like Ice Technology? This is to say what our core founding team and CEO Dr. Gu Lingyun did in the US at the time. In the United States, our core team uses data from non-bank systems, such as behavior, social, e-commerce, communications and other data to build an assessment analysis model to serve long-tail consumers and small and micro enterprises, providing them with financial services. service. This group of people accounts for about 15% of the US population. In China, there are a wider group of the same group, including small and micro enterprises, farmers, blue collars, etc., accounting for about 60%-80% of the population. These people are very vulnerable when faced with traditional financial institutions, such as banks. Because these people are relatively undocumented or have very thin records in the banking system, and there are not a lot of fixed assets, banks are not willing to go, or have no ability to serve them. These people have both financial needs and the availability of such financial services to them. Helping financial institutions to provide financial services to these people is the initial heart of Ice Jian Technology. When ice is helping financial institutions to serve these long-tailed people, there are mainly several keywords. The first one is called small amount, the second is called high frequency, and the third is called automation. When traditional financial institutions provide credit business, they have good mortgages and car loans, and the quotas are relatively large, such as more than one million. The personal financial needs are more than a few thousand, tens of thousands, and hundreds of thousands, and basically no more than one million. The small amount of financial demand represents a very high frequency of this scene. This is good for data and algorithms. Only in such a scenario can there be enough samples to train in order to realize its value through the model. Financial institutions have a large number of small high-frequency samples.对于这些金融机构,他们面临一些问题:第一个是人员方面的问题。比如各种农商行和城商行,这是银行体系内相比于五大行或者股份制银行,更愿意做下沉的服务,去服务于这些人群和机构。对于这样的机构来讲,他们面临的一大问题在于人才的缺失。他很想分析现在业务中积累的数据,但是数据分析的人才,特别是风控领域,相对来说高端的人才来源是非常有限的。这些城商行和农商行不在一二线城市,积累人才过程中,有很大的劣势,特别是核心的风控方面的数据分析人才。即使能够给出同样的薪资,百万或者千万,也不能够有足够的竞争力吸引到这样的人才。因为这样的人才,更多愿意留在大城市。这样的机构,对我们是有需求的。第二个是业务方面,传统银行提供金融服务的时候,依赖的是客户经理,需要线下通过客户经理一个一个审核个人和小微企业,才能做出评估,线上服务处于劣势。相比于银行和传统金融机构,现在涌现出来的互联网金融机构反而在线上服务于这些个人和小微企业方面比较领先,比如大家都耳熟能详的一些产品,像花呗、借呗、微粒贷,或包括像趣店分期乐的消费分期产品,可以看出走在前面的反而是非银行机构。所以,对于我们这样的第三方机构,通过服务这些非银机构,可以积攒一些知识,获得进步。同时在服务各种传统金融机构的同时,也可以把学习到的知识传递给他们,正如今天的主题是超级进化者,也是在帮助他们进化。通过服务先进互联网机构和传统金融机构,帮助银行把线下挪到线上去。数据在我们服务的过程中也很重要,我们需要在高频场景下获得大量的样本,再去训练这样的模型。我们获得的样本,更多是模型需要预测的变量,叫目标变量。但是模型还有其他的变量,在评估个人和小微企业的时候,到底基于什么样的数据去评估这样的个人和小微企业呢?这要说到移动互联网在中国的优势,大家可以看到西方的发达资本主义国家,美国也好,西欧也好,它的金融机构经历了上百年的发展历史。在支付方面习惯信用卡和支票体系,但这些在中国还没落地之前,移动互联网以后来者居上的姿态,占领了这个位置。我们国家非金融机构、个人和小微企业积累的数据非常丰富,不管是电商、消费还是通讯、地理位置,很多都集中在移动互联网,集中在手机当中。这些数据在普惠金融的场景下,都可以得到很好的应用。所以,我们作为一家第三方机构,一方面是通过跟金融机构的合作,拿到大量样本所需要的目标变量;一方面去整合移动互联网场景,帮助小微企业和个人进行信用评估。当然,在做这个工作过程中我们也感受到,当我们的生活越便捷、越多的数据去沉淀,获得这样的附加增值服务的时候,隐私也更加缺乏,这点大家应该也有很深的体会。隐私和便捷获取数据这两点之间,是很难兼得的。但是,在这一过程中,冰鉴科技力求合规使用数据。我们向金融机构输出的,是通过模型高度加工的评分,而不是各种隐私的评分。上面说的是移动互联网,因为中国在这方面走在前沿,所以在普惠金融方面也积攒了更多的数据。除了数据,在我们为金融机构服务过程中另一个重要的东西就是我们的AI算法,包括自然语言处理、知识图谱、机器学习、深度学习。我们要通过这些技术对算法模型和模型效果进行优化。比如,在小微企业的评估当中,除了将一些非公开的数据,进行脱敏输出之外,也需要对一些公开数据,包括一些舆情通过自然语言算法进行处理,便于实施对小微企业的一些监控。又比如关联知识图谱,这是风控领域的下一个方向。In the past, the risk assessment for individuals and small and micro enterprises was more to evaluate individual individuals and enterprises, while the use of associated knowledge maps was to consider different individuals and companies in a network relationship. Since people and businesses will live in a social network in the future, this includes the company’s guarantee network, including social behavior networks and communication networks. The relevant one or two layers in these networks have their influence on the credit status of people on a single point. By associating a knowledge map, when evaluating a person, evaluating the relationship between one and two layers of his association can ultimately lead to more accurate results. For Inclusive Finance, as a third-party service organization, we can only do our best to create more value with artificial intelligence technology and data. But the path of inclusive finance is very long, and it also includes the structure of a series of financial institutions’ infrastructure, including the gradual standardization of regulation in this industry. I hope that Ice Jian will do his part in this middle and hope to see that the inclusive financial industry can develop better and better. Click on the video above to see more.

Tags : Internet entrepreneurshipInternet entrepreneurship projectStart a business