Javascript is everywhere, from enterprise to personal projects, enhanced by superset languages such as Typescript and ReasonML. Despite being the third most popular programming language in Github for Machine Learning, most data scientists and companies prefer Python as their language of choice. Javascript Machine Learning has seen a leap of growth in 2018, although many notable projects are still being unmaintained, many key players including Brain.js and Tensorflow.js have been evolving throughout the year of 2018 and up until now of 2019.

Here is the list of best Javascript/Typescript Machine Learning libraries in 2019 that qualifies the following criteria

• Maintained and has an active community
• Production ready
• Popularity

Disclaimer: Although the title says the best “Machine Learning Javascript libraries”, I’ve included libraries that are not strictly ML but have a tight relationship.

1. Tensorflow.js

Tensorflow.js in 2019 has become the bread and butter for all Machine Learning Javascript projects due to its comprehensive linear algebra core and deep learning layers. It has rapidly caught up with its Python sister in the number of supported APIs and almost any problems in Machine Learning can be solved using it at this point.

In addition to providing Deep Learning and Machine Learning in Node.js environments, Tensorflow.js can be used directly in the browsers, while leveraging WebGL for accelerations. The Tensorflow.js model of supporting both browsers and Node.js environments has been adopted by many open source libraries including brain.js and machinelearn.js.

2. Brain.js

With an aim to reduce the complexity of machine learning and deep learning into something that a child can understand, Brain.js certainly qualifies as the easiest deep learning library out in the wild including the ones in the Python ecosystem.

Being one of the most actively evolving Javascript Deep Learning library, they’ve recently launched a fun new interactive tutorial:

https://scrimba.com/g/gneuralnetworks

Also, I recommend Brain.js to any newcomers to Deep Learning since you can find a range of terrific tutorials online that are aimed at absolute beginners of Brain.js and Deep Learning itself.

Check out one of the most viewed Brain.js tutorials on Youtube:

Also, make sure to take a look at their Github repository and the website.

3. stdlib-js

Plotting and sample datasets are essential components of any Machine Learning workflow. Plotting provides a deeper insight into the dataset and assists the Machine Learning engineer to decide the next action in order to solve the problems in hands. Furthermore, the library is equipped with numerous sample datasets of different types including binary classification, regression and suite of image datasets.

The library can be a great companion to any machine learning project as it provides a handful of essential machine learning operations as mentioned above and has great documentation.

Check out their Github repository and website for further information!

4. machinelearn.js

Machinelearn.js is a new kid on the block to fulfil the role that ScikitLearn does in the Python land for Javascript Machine Learning. It provides straightforward and mission-critical models and utilities for supervised and unsupervised problems. Focusing on the simplicity and all-in-one general purpose machine learning for Javascript and Typescript developers, it provides clustering, decomposition, ensemble, bagging, linear models, feature extractions and more.

It leverages the battle-tested library Tensorflow.js for its core math, which enables acceleration using C++ native interface, CUDA and WebGL. Unlike its peer mljs, machinelearn.js is written in Typescript, which provides type definitions for users who are already using the language.

Check out their Github repository and website.

5. Math.js

It’s a library for all math needs in Javascript with extensive linear algebra APIs including matrix operations and core math. It is quite lightweight as it does not depend on other acceleration techniques such as WebAssembly or WebGL.

경축! 아무것도 안하여 에스천사게임즈가 새로운 모습으로 재오픈 하였습니다.
어린이용이며, 설치가 필요없는 브라우저 게임입니다.
https://s1004games.com

With a large number of contributors from the community, math.js has one of the most vibrant open source community managed by core developers including Jos de Jong.

In the domain of Javascript Machine Learning, Math.js is sufficient enough to fulfil the role of Numpy as it provides all the essential utilities that are required to solve machine learning problems.

Check out their website and Github repository today.

6. face-api.js

Face-api.js includes implementations of well-known Face Detection and Recognition models pre-trained on a wide variety of datasets. It gives you a ready-to-use APIs that you can directly plug into any Node.js and browser environments.

The library implements SSD MobileNet V1,Tiny Face Detector and MTCNN for Detection architecture, and an architecture similar to ResNet-34 for face recognition. Being lightweight using Tensorflow.js, the library can be used on both mobile and web browsers with no issues.

Check out the Github repository and the author’s blog article.

7. R-js

If you are looking for a hardcore math utility like in R, a member of the community has ported (almost) the important components of R into Typescript.

Linear algebra specification such as BLAS is widely adopted by the heart of machine learning applications such as Numpy and R. The project R-js has re-written the entire BLAS into Typescript, which sets the foundation for future projects to implement more powerful math base framework like Numpy and SciPy of Python.

Check out their Github organization.

8. natural

If you are looking for an NLP tool like NLTK or spaCy in Javascript, natural is currently the most well maintained and active NLP companion library to the date.

It provides essential APIs required for machine learning NLP workflow including preprocessing utils including Tokenizers and Stemmers, and searching utility like tf-idf.

It has a built-in sentiment analysis using a simple algorithm a handful of data structures including Tries, ShortestPathTree and LongestPathTree.

Check out their Github repository.

[출처] https://towardsdatascience.com/top-javascript-machine-learning-libraries-in-2019-cb63b95bdd10