Evaluation of ML methods for online use in the browser
Hao Liu · Figshare · 2021
Machine learning continues to be an increasingly integral component of our lives, whether we are applying the techniques to research or business problems. Machine learning models ought to be able to give accurate predictions in order to create real value for a given organization. At the same time Machine learning training by running algorithms on the browser has gradually become a trend. As the closest link to users in the Internet, the web front-end can also create a better experience for our users through AI capabilities. This article will focus on how to evaluate machine learning algorithms and deploy machine learning models in the browser.We will use "Cars", "MNIST" and "Cifar-10" datasets to test LeNet, AlexNet, GoogLeNet and ResNet models in the browser. On the other hand, we will also test emerging lightweight models such as MobileNet. By trying, comparing, and comprehensively evaluating regression and classification tasks, we can summarize some excellent methods/models and experiences suitable for machine learning in the browser.