/sci/ - Science & Math » Thread #11630113

10KiB, 235x238, 4ea2b745af3ed62141895496fba3a49f.jpg

View Same Google iqdb SauceNAO

Machine learning is DEAD

Anonymous Sat 02 May 23:15:35 2020 No.11630113 View Reply Original Report

Quoted By: >>11630140 >>11630162 >>11630170 >>11631769

>What a time to be alive
>ML = AI
>Strong AI
>Full self-driving = adding software patches™ to a dead end infrastructure for a price®
>Instagram filters
>Chess player
>Popsci nuclear fuel
>Kurtzweil
>Singularity™
>Hallucinating reality

Now that the machine learning field completely full of shit and dead, what's next for humanity in order to achieve AI?

Anonymous

Anonymous Sat 02 May 2020 23:22:08 No.11630135 Report

Quoted By: >>11630161 >>11630262 >>11632035

I came up with a revolutionary idea which makes strong AI possible.
But what is the best way to publish it?

Anonymous

Anonymous Sat 02 May 2020 23:22:55 No.11630140 Report

Quoted By: >>11631964

>>11630113
Machine learning has been around for decades in the form of pattern classification theory. It's still used all the time, e.g. in classifying satellite images. Why did people just recently latch onto it and start touting it as AI?

Anonymous

Anonymous Sat 02 May 2020 23:29:09 No.11630161 Report

Quoted By:

>>11630135
Post it here.

Anonymous

Anonymous Sat 02 May 2020 23:29:10 No.11630162 Report

Quoted By: >>11630168 >>11630179

>>11630113
We need new hardware, a way to replicate how neurons function physically instead of with software.

Anonymous

Anonymous Sat 02 May 2020 23:32:15 No.11630168 Report

Quoted By: >>11630180

>>11630162
just open a farm of autistic kids

Anonymous

Anonymous Sat 02 May 2020 23:32:41 No.11630170 Report

Quoted By: >>11630216 >>11631695

>>11630113
Is this bait? How is driving on roads a dead end infrastructure?

Anonymous

Anonymous Sat 02 May 2020 23:37:10 No.11630179 Report

Quoted By: >>11630187

>>11630162
We don't need new hardware to achieve true AI because software can emulate it, although it might be 100 times slower than us.

Anonymous

Anonymous Sat 02 May 2020 23:37:14 No.11630180 Report

Quoted By:

>>11630168
Autists don't breed well, a lot of them are asexual, they're human pandas

Anonymous

Anonymous Sat 02 May 2020 23:38:29 No.11630187 Report

Quoted By:

>>11630179
youre an idiot

Anonymous

Anonymous Sat 02 May 2020 23:48:21 No.11630216 Report

Quoted By: >>11630243 >>11631695

>>11630170
Tesla driving software-hardware is reaching a point of diminishing returns, and no matter how hard they train those cars, they are not progressing significantly anymore. There is a reason why """FSB""" is being offered in multiple "convenient pay to use" small packages: FSB will never, EVER happen.

Anonymous

Anonymous Sun 03 May 2020 00:00:13 No.11630243 Report

Quoted By: >>11630245 >>11630278 >>11631695

>>11630216
Source on the research? How do we know it’s slowing down?

Anonymous

Anonymous Sun 03 May 2020 00:01:23 No.11630245 Report

Quoted By: >>11630278

>>11630243
It's probably a shitpost

Anonymous

Anonymous Sun 03 May 2020 00:13:06 No.11630262 Report

Quoted By:

>>11630135
Don't post it anywhere. You'll get assassinated for thirdage information. No joke.

Anonymous

Anonymous Sun 03 May 2020 00:19:43 No.11630278 Report

Quoted By: >>11631695

>>11630243
>>11630245
Ask Tesla buyers/users lmao

Don't forget that Tesla has been training those things since 2012, there is not a lot of room for improvement at this point

Anonymous

Anonymous Sun 03 May 2020 09:55:09 No.11631695 Report

Quoted By: >>11631964

>>11630170
>>11630216
>>11630243
>>11630278

Just because Tesla's doing it wrong doesn't mean it's impossible

https://www.youtube.com/watch?v=aqdYbwY9vPU

Anonymous

Anonymous Sun 03 May 2020 10:21:23 No.11631769 Report

Quoted By:

>>11630113
>Now that the machine learning field completely full of shit and dead

>been working on a betting/gambling system over the last couple of years part time(not telling you anything other than that it's based on a sport)
>most of the work is around data pipeline, understanding the data and IRL event you are modelling
>tfw is profitable (currently returning 10-15%)
>tfw gambling winnings aren't taxed in my country

Sure talk shit nigger but this is on track to buy me a house in 6 months

Anonymous

Anonymous Sun 03 May 2020 12:10:40 No.11631964 Report

Quoted By:

>>11631695
He is obviously an arrogant megalomaniac who merely implemented the basic functions of self-driving which in this day probably any smart EE/CS grad could do.
The hard thing about self-driving right now is getting it to work in 100% of the cases instead of just 80-99%. You can't really make a machine learning system learn all the edge cases by itself, because there is too little data.

>>11630140
It's gotten a lot easier to use and more performant when people figured out how to effectively train "deep" neural networks a few years ago.

Anonymous

Anonymous Sun 03 May 2020 12:36:57 No.11632035 Report

Quoted By:

>>11630135
Knowledge bomb it on an audio hard track and millions of usb/sd cards.

Capcode	All Only User Posts Only Moderator Posts Only Admin Posts Only Developer Posts
Show Posts	All Only With Images Only Without Images
Deleted Posts	All Only Deleted Posts Only Non-Deleted Posts
Ghost Posts	All Only Ghost Posts Only Non-Ghost Posts
Post Type	All Only Sticky Threads Only Opening Posts Only Reply Posts
Results	All Grouped By Threads
Order	Latest Posts First Oldest Posts First

Your latest searches