Research On The Key Of Smart Finance

In recent years,with the rapid development of Internet,artificial intelligence,ma-chine learning and other information technologies,financial technology has become the most popular topic in the financial industry.In the era of financial technology,more and more financial services will gradually be digitized,standardized,and intelligent.When it reaches a certain stage,the existing technical bottleneck of financial institutions will no longer exist.The future competition of financial institutions will depend to a large extent on the customization of personalized services.This subject will be the future stage?it does not take a long time?called smart finance,where smart investment related to wealth appreciation is of the greatest concern.Robo-advisor refers to a series of factors based on the customer's own financial needs,asset conditions,risk tolerance,risk appetite and other factors.It uses modern investment portfolio theory and artificial intelligence tech-nology to build a data model and build a network platform.Investors provide intelligent financial consulting services,which can replace traditional manual investment consul-tants to some extent.Simply put,smart investment is the use of artificial intelligence technology,the use of portfolio theory,for the user to develop a matching investment portfolio.With people's concern for wealth appreciation,financial institutions such as commercial banks,fund companies,and asset management companies have all launched smart investment services.In the near future,robo-advisor will become an important tool for residents'asset allocation.In recent years,many financial institutions and research teams are deeply researching the development trend of smart finance and the prospects of robo-advisor and technology development.Its purpose is to accelerate the popularization of artificial intelligence,ma-chine learning and other technologies in the financial sector.By solving problems such as financial time series analysis,investment portfolio,and personalized recommendation,the expected return rate of the intelligent investment model can be further improved.Under this background,the subject first designed a set of intelligent financial architecture based on machine learning algorithms,and then conducted in-depth research on key technolo-gies such as financial time series analysis,personalized recommendation,and intelligent investment advice in smart finance.It puts forward ideas such as robo-advisor for group,financial social networks,and real-time transaction feedback.The main contributions and innovations of this work are summarized as follows:1.In-depth study of the impact of rapid development of information technology on tra- ditional commercial banks,from the growth rate of Internet financial market share analysis of the impact of traditional commercial banks.In order to cope with chal- lenges and opportunities,a brand-new intelligent financial architecture was proposed from the perspective of machine learning,and on this basis,the research content and implementation path of financial time series analysis,personalized rec- ommendation,and intelligent investment advice were proposed.Financial time series forecasting helps to improve the accuracy and generalization performance of intelligent investment screening,accurate analysis of customer portraits,and per- sonalized recommendations.It is conducive to accurate push of intelligent invest- ment advice.This article first analyzes the market potential of smart investment and investment from the changes in the basic customer base of the bank,and then demonstrates the structure of the intelligent financial system around products,chan- nels,and users.Finally,from the perspective of machine learning,it elaborates the need to achieve smart financial needs.Based on this work,we have proposed key technologies such as financial time series analysis,personalized recommendation, and intelligent investment advice from the practical issues faced by commercial banks,which provide the basis for financial practitioners and information technol- ogy researchers.2.In-depth study of the characteristics of time series in smart finance and analysis and prediction methods,a novel algorithm based on random Fourier maps( l_2,smart1RF- ELM)for financial time series prediction algorithm.Financial time series fore- casting is a complex task because the behavior of investors is affected by many small and unpredictable factors.In this paper,in order to maximize the return on investment and effectively manage the liquidity risk,we use an over-limit learning machine based on the l_2,1norm and random Fourier mapping( l_2,1RF-ELM)is applied to financial time series forecasting problems.Compared with traditional neural network algorithms,the advantages of ELM in efficiency and generalization performance have been proven in a wide range of problems from different fields,thanks to the integration of the l_2,1norm, l_2,1RF-ELM is able to automatically prune irrelevant and redundant hidden neurons to form a more discriminating and compact hidden layer.Comparing the performance of l_2,1RF-ELM with other machine learning algorithms,the effectiveness of our algorithm is verified in a given data set.3.In-depth study of key technologies and processing methods for personalized recommendation in smart finance,and a two-way evolution based financial product recommendation model.In the practical application of the financial sector,the personalized recommendation system for financial products usually takes long-term wealth appreciation as its effectiveness because it cannot be realized immediately. At the same time,the recommendation of financial products also depends on external factors such as market conditions and government supervision.In this paper,considering the dynamic changes in investor risk tolerance and investment preferences,we propose a bidirectional evolutionary financial product recommendation system?BDE_RS?to solve this problem.This algorithm attempts to balance the short-term of the recommendation system.Accuracy and long-term benefits.Extensive experiments on benchmarks and actual data sets show that the proposed method outperforms other given machine learning algorithms.4.In-depth study of the key technologies and processing methods of asset allocation in smart finance,and propose a financial product recommendation algorithm based on incremental extreme learning machine?IELM?.Because of the economic environment,market behavior,and transaction data are constantly changing,traditional recommendation systems cannot be directly applied to asset allocation.Combining the characteristics of initial data sparsity,economic environment dynamics,and transaction behavior continuity,a new financial product recommendation system was proposed based on the optimization of the incremental learning machine ?IELM?.The main idea of this algorithm is to start learning a few neurons.Considering changes in the behavior of investors and asset allocation,new neurons are constantly added until the entire process is stable.Among the given data sets,the IELM algorithm is compared with other incremental algorithms in detail.All the results show that the proposed IELM algorithm has better performance than other state-of-the-art methods.5.In-depth study of the key technologies and solutions for smart investment in smart finance,and an novel robo-advisor algorithm based on incremental multi-kernel extreme learning machine is proposed.Smart investing is a type of robot based on the needs of investors for financial management.Through machine learning algorithms and products,financial consultant services that rely on human intervention in the past have been completed.However,many previous general algorithms are not suitable for information fusion in heterogeneous data in financial time series.We propose an incremental multi-core learning machine?IMK-ELM?model that initializes a general training database and then dynamically adjusts to the classification task.The IMK-ELM algorithm proposed in this topic can update the training data set and adjust the weight of multiple information sources at the same time.Among the given data sets,simulation experiments show that the algorithm proposed in this topic far exceeds the performance of the Shanghai Composite Index over the same period in terms of investment performance.6.In-depth study of the key technologies and processing methods of user clustering in smart finance,and propose an novel robo-advisor algorithm based on user investment preference clustering.Intelligent investment advice is a new type of financial advisor model,which can provide investors with financial consulting and investment management online.Data clustering and project recommendation are very important and challenging in the intelligent investment and care model.These two tasks are usually considered independently.However,the similarity calculation in project recommendation has a strong correlation with user clustering.For example,a large number of financial transactions include not only the user's asset information but also the user's social information.The existence of relationships between users prompts us to cluster and at the same time we can optimize project recommendations in smart investment advice.In this article,we provide a guideline that captures the relationships between users and groups,and proposes a brand new smart advisor framework.Among the given data sets,simulation experimentsshow that our proposed framework outperforms given machine learning algorithms in both performance and investment performance.