A powerful and easy-to-use machine learning platform is very important for machine learning research . A good machine learning framework will provide a wealth of components , It can facilitate the design and implementation of machine learning model .
At present, there are several basic machine learning platforms ：
Open source machine learning platform ,API（Application Programming Interface, Application programming interface ） Rich and free , But learning costs are high , for example R、Python、Mahout、Spark MLlib etc. . Commercial machine learning platform , Such platform algorithms are limited , But it has been tested by long-term practice , There are few system problems , The cost of learning is low , Little or no programming , But the analysis model in the system is not rich , for example IBM SPSS Modeler. Graphical machine learning platform , This kind of platform combines the advantages of the above two types of platforms , It provides rich algorithm call interfaces , It can also quickly build the workflow of machine learning through the graphical human-machine interface , And it can reduce the workload of programming . Currently Intel 、 Microsoft 、 Google and domestic BAT（ Baidu 、 Alibaba 、 tencent ） And other companies have provided such machine learning platforms .Caffe2 
Caffe2 It is a framework for industrial applications , Widely applied . But from the perspective of installation and deployment ,Caffe2 User experience Not very friendly , Official documentation and tutorial support are not sufficient . and Caffe2 Only support Python 2, This limits its future expansion .
MXNet
