>>53818552
no i will not install gentoo and im not from pol

>>53818541
>neural networks seem to be mainly for numbers and not words
why do idiots keep spouting this?

>>53818568
you do realise NNs work with continuous variables, not discrete ones, right?

>>53818568
are you serious? how do I know you're not an idiot? I'm pretty sure neural networks are not generally used to work with words because it is hard to standardize/normalize/convert/whatever the words into numbers. I mean I guess you could somehow but....I dunno...I can't tell if you are stupid or I am

>>53818602
I installed arch because I live on the fucking bleeding edge which you gentoo faggots can't even handle

just kidding, but I did install arch and i riced it so yeah, i'm a faggot

but we need to focus on AI and chatbots

>>53818618
I literally have not been on pol more than 5 times in my life and based off your response I'm pretty sure you are stupid and don't know a fucking thing about neural networks

>>53818641
Listen here, Barry Rasmussen, I am not autistic.

>>53818663
Listen fucking here Lindsay RAndolph, I don't care how many times you change your goddam name or where you work but I will not install gentoo unless you fucking ask me to . DO YOU UNDERSTAND ME LINDSAY FUCKING RANDOLPH OR BARRY FUCKING RASMUSSEN OR WHATEVER YOUR STUPID NAME IS? I WORK MY TAIL OFF TO RAISE 9 SMALL AFRICAN AMERICAN BOYS AND I WILL NOT TAKE THIS UTTER BULLSHIT FROM YOU. I WILL NEVER INSTALL GENTOO. GET THAT THROUGH YOUR TINY LITTLE FUCKING SKULL. but i also think you are lying but that is what neural networks are...........hmmmmm....im 19 and just made one for fun as well....i used a tutorial though....anyways...have a good day

>>53818663
but wait a second if you're so good at neural networks, why would you say that idiots say that words are not good for neural networks? literally everywhere on the internet says words and neural networks are not good together

like this person (Blanche Vaughn)
>>53818593

i think you are lying desu

>>53818541
>neural networks seem to be mainly for numbers and not words
No. Absolutely fucking no. MNIST is mainly used for finding areas for improvement and testing. Neural networks can be applied to far more than that.

Read the docs for TensorFlow, Google even mentions some information for how they do search predictions - it's not just a simple "take the phrase and then find out what others have searched for containing that phrase", it does a lot of analysis to find out the context of what you're typing, which involves looking in to the position of each word to the former and latter words, combined words, etc.

I'd just give up, developing significant and true AI (the term AI is misused everywhere) is only possible if you work at Google, IBM and a few other companies or if you're extremely skilled.

>>53818813
Looked into TensorFlow, looks awesome. Thanks.

I'm not trying to develop a true AI, really just a chatbot

>>53818568
100% this

http://karpathy.github.io/2015/05/21/rnn-effectiveness/

Given enough training data, a LSTM could make a perfect chatbot.

>>53818541
https://github.com/Microsoft/BotBuilder

>>53818593
Are you fucking daft Blanche? You clearly have no fucking clue how neural networks work. Internally they use continuous values but they can easily take inputs from a discrete space, as long as that space lies within a continuous space.

>>53818541
>I'm pretty average at math and don't really like algorithms
>neural networks seem to be mainly for numbers and not words
You're retarded. Give up.

i've decided that i am retarded and am going to give up