Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
MLHelper: Algorithms and utils for Machine Learning in JavaScript (github.com/laoqiren)
65 points by laoqiren on Nov 18, 2017 | hide | past | favorite | 7 comments


I want to point you to https://github.com/stdlib-js/stdlib. You can use it for the lower level utils. There are almost more matrix utils in the js world then there are devs. Let's stand each others shoulders.

Let's standardize a little more than what lodash/underscore does.


I just don’t understand: why? Beyond the novelty, why are people spending time doing this in JS?

I code things in JS because I spend a lot of time with JS. So a quick CLI tool others would write in Go, I might do in JS just because it’s so familiar. But I also spend a lot of time in Java. I’d never choose JS over Java for something like this. I feel like the only reason you would ever do this in JavaScript is because you literally don’t know any other language.


There's a huge demand for explanatory material for ML, and running small-sized demos in the browser is a great tool for that purpose.

I don't know how many people are using JS to train particularly large models, or models they'd use in production, however.

Finally, I wouldn't be surprised if we see something like "ML tools for Brainfuck" sometime on HN, because everyone wants to learn ML and implementation is a good way to learn and everyone wants to write something new and "fun" (for various definitions of fun).


Yes, making some wheels in your language is a good way to learn


Can anyone comment on which JavaScript ML libraries are most popular or most commonly used? I've just started using ml.js and it does what I need, but I feel like the ecosystem is so huge I could be missing out on some real gems


Thanks for your votes!


Could be interesting to see something like this integrated with DeepLearning.js




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: