Design And Implementation Of Web-based Machine Learning Visualization System

People are deluged with a significant amount of diverse information due to the advent of the big data area.According to statistics,more than 80% of the data human obtained comes from the visual system.A good visualization system can effectively help people to get potentially useful information and knowledge from this data.Furthermore,with the development of technology,the data models become more and more complex,and it is necessary to combine visualization technology with machine learning technology to process data more efficiently.However,most researchers focus on the optimization of the machine learning algorithm itself,but they take no count of the visualization of the execution processes or the results.This paper analyzes three visualization fields related to machine learning technology and finds that there are few machine learning visualization tools are available.To solve this problem,the specific work is as follows:1.This paper proposes a Web-based visualization method for machine learning by combining machine learning technology with Web front-end D3.js visualization technology.Applying jQuery,JavaScript,Bootstrap,and some other Web front-end technologies,the new method achieves the visualization of the k-nearest neighbor algorithm(KNN)and the support vector machine(SVM)algorithm.It also supports the timely adjustment of the algorithm parameters in the front-end.2.This paper designs and implements a web-based machine learning visualization system,and gave the overall design of the machine learning visualization system.There are five modules in this scheme,including the data management module,the visualization algorithm implementation module,the interface design module,the D3.js drawing module and algorithm evaluation module.The data management module is further subdivided into three submodules: data generation,data import,and the data export module.The visualization algorithm module includes the KNN classification visualization algorithm,KNN regressionvisualization algorithm and SVM visualization algorithm module,detailed design and implementation of each module's content.This system can directly adjust the parameters of the machine learning model through the Web front-end visualization technology.It is useful in practice.3.Finally,the proposed system is implemented and the relevant data sets were imported for the test.The experimental result shows that the proposed web-based machine learning visualization system is feasible and effective.