Web Saturation with Libraries of Machin Learning Modules

Abstract

The paper presents the result of Web resources cursory review (45 sources, through to December 2019) containing software libraries that implement models and algorithms of artificial intelligence (Machine Learning - ML). The purpose of the review is to provide a set of developments in the field of ML for students and developers of digital technologies in all fields of activities compactly. The ML Libraries (MLL) have been classifying by languages implementation and by basic methods specificity. The table “MLL-Features” has been compiling. The features characterize a set of methods, a technique and a task orientation of each library. In total, there are more than 40 such libraries for developing server-side and client-side of knowledge-based Information Systems (KBIS), deep Artificial Neuron Network (ANN), and Natural Language Processing. Most of them are new but are ready for use by developers. As the review one result, it became clear that ML libraries are quite accessible to students and even high school students. The advantage of Web languages is that they are useful both for understanding ready projects and for direct use: only a web browser and development environment are required to run an ML project in JavaScript. The review is useful for undergraduates, graduates, postgraduates and novice developers of artificial intelligence systems and KBIS.