[Chicago] Writing fizzbuzz for the new century

[JS Irick]

This is actually pretty interesting to think about.

1)  Are Tensorflow and related libraries actually probabilistic like I'm
assuming?  My uneducated abstraction is that they iterate over the data set
using some sort of randomization to create "rules" then apply them.
2)  If so, how do you test them?  Do you do multiple iterations of a test
and look at the distribution?
3)  How do you avoid "local minima", so to speak.  I saw a "step size"
reference in the article, is this how much the attempted algorithms change
each time?

Thanks!
JS

On Wed, May 25, 2016 at 11:13 PM, JS Irick <hundredpercentjuice at gmail.com>
wrote:

> I can't tell if you're trolling or not.
>
> He should have written unit tests for his machine learning program?  On
> the plus side, that's one the best definition of "results oriented" I've
> heard.
>
> On Wed, May 25, 2016 at 10:46 PM, Michael Tamillow <
> mikaeltamillow96 at gmail.com> wrote:
>
>> He should have wrote some unit tests first!
>>
>> On Wed, May 25, 2016 at 5:29 PM, Joshua Herman <zitterbewegung at gmail.com>
>> wrote:
>>
>>> I saw that on Hn . The funniest thing is that at the end the trained
>>> tensor flow network doesn't even return the correct values after it is
>>> trained.
>>> On Wed, May 25, 2016 at 5:09 PM Robare, Phillip (TEKSystems) <
>>> proba at allstate.com> wrote:
>>>
>>>> Joel Grus (author of O'Reilly's "Data Science from Scratch: First
>>>> Principles with Python") has written a hilarious post about being asked to
>>>> write FizzBuzz in a Python interview.  So he sets up a neural network.
>>>> Check out http://joelgrus.com/2016/05/23/fizz-buzz-in-tensorflow/
>>>>
>>>> Of course, as one of the comments points out, if you have to do
>>>> fizzbuzz in Java there is the Enterprise version of FizzBuzz with all the
>>>> GoF patterns:
>>>> https://github.com/EnterpriseQualityCoding/FizzBuzzEnterpriseEdition.
>>>>
>>>>
>>>> Phil Robare
>>>> _______________________________________________
>>>> Chicago mailing list
>>>> Chicago at python.org
>>>> https://mail.python.org/mailman/listinfo/chicago
>>>>
>>>
>>> _______________________________________________
>>> Chicago mailing list
>>> Chicago at python.org
>>> https://mail.python.org/mailman/listinfo/chicago
>>>
>>>
>>
>> _______________________________________________
>> Chicago mailing list
>> Chicago at python.org
>> https://mail.python.org/mailman/listinfo/chicago
>>
>>
>
>
> --
> ====
> JS Irick
> 312-307-8904
> Consultant: truqua.com
> Coach: atlascrossfit.com
> Programmer: juicetux.com
>



-- 
====
JS Irick
312-307-8904
Consultant: truqua.com
Coach: atlascrossfit.com
Programmer: juicetux.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/chicago/attachments/20160525/1fdc2d6b/attachment.html>