[Jprogramming] JHS update

[greg heil]

Joe

>i of course, as most folk do not care about the data. i am only interested in my data of the day:-) What is interesting to me is the TOOL, and how it can be applied to ANY data. Google and most tech companies are pushing ML now. i found their ~playground tool fascinating and useful. If that could be ported (and generalized?) to JHS it would be wondrously useful and attracting.

---~
http://playground.tensorflow.org/

greg
~krsnadas.org

--

from: Joe Bogner <joebogner at gmail.com>
to: programming at jsoftware.com
date: 20 October 2016 at 17:46
subject: Re: [Jprogramming] JHS update

greg, yes, lots of data. The trouble is selecting something that would be
interesting to most people

here are some other resources:

https://github.com/caesar0301/awesome-public-datasets
https://github.com/curran/data

--

from: greg heil <gheil.j at gmail.com>
to: Programming forum <programming at jsoftware.com>
date: 20 October 2016 at 16:34
subject: Re: [Jprogramming] JHS update

There is a LOT of ~publicdata!-)

There is also a LOT of encouragement to explore it using ~machine-learning.

Brian (maybe others?) has made a shot at ML in ~feedforwardNN.

Making good NN's though is very much a finickey, interacting with data, thingee.

An example, with many stubs, "Hello World"s for Jd, ML...

Google gives a (Github'd) "~playground" example, that might be useful (in JS).

---~
https://www.google.com/publicdata/directory
http://www.infoworld.com/article/3131515/artificial-intelligence/4-google-data-sets-to-kickstart-machine-learning.html
http://code.jsoftware.com/wiki/User:Brian_Schott/code/feedforwardNN
http://playground.tensorflow.org/

greg
~krsnadas.org