Machine Learning Algorithms in a nutshell

Machine Learning Algorithms in a nutshell

it overfitted, back to start

Teaching old people how to use a computer in a nutshell.

Yup, except after the machine learns the most efficient way to do something, it can network with every other machine like itself and just teach them what it learned and visa versa. Imagine if you only had to teach one grandma how to use a computer, and suddenly every grandma knew how to use a computer.

But what if that first grandma settled on a locally optimized version of using a computer that auto-accepted any and all requests to install toolbars?

Reminds me of the google dude walking. I figure eventually if you run the program enough times it'll resort to walking on its hands instead. That would be entertaining.

Old people are basically AI anyways

Old people should be interred in huge mechanical war machines called dreadnoughts so they can better serve the emperor, even in death.

Unfortunately, it's unlikely given its initial start parameters (oriented with legs down in bipedal stance). Once it learns moving forward with its legs it would never start unlearn it and start using its hands.

It'd be hilarious though

Officially renaming all my essential correction functions to warmer(), colder(), and youreOnFire()

Descend that gradient!

A controversial opinion, to be sure, but I respect your right to hold it.

Aw it got cut off at the best part. "AHHH IT BURNS!"

https://www.youtube.com/watch?v=KNZSXnrbs_k
http://i.imgur.com/gE7CDx0.jpg

That convergence though

I'd assume the reward (moving forward) would plateau and the neural network would adjust weights accordingly so that it would less likely to fall. So one of the intermediary goals would be to never fall.

Correct me if I'm wrong, but I'm did you just prove all grandmothers are geth?

Huge, if true.

So how do you teach it to stand up when it falls? Oh wait that is not on the list of things the A.I. has to be trained for, so just assume it never happens.

I work on reinforcement learning, and I can confirm that this is an accurate representation of my experiences with machine learning, only Patrick learns faster.

And thus the cookie apocalypse begins!

Machine Learning Algorithm making a peanut butter jelly

you forgot the part in ur comment where you refer to yourself as an intellectual

Learning rate too high

Come on, guy. You're in the comments of /sub/videos telling random people to read a 115 page scholarly essay on Native Set Theory because it's a 'Pro-Tip'. What the fuck are you doing

A surprise, to be sure, but a welcome one.

No you have a separate AI whose job that is

Right, but essentially every episode was the same. Go with Road Runner. Coyote comes up with crazy invention to catch road runner, invention backfires, and Coyote gets hurt. Wash, rinse, repeat. Now I absolutely love Looney Toons so I'm not complaining about this. But Spongebob just does this in a different way. Was the joke in the clip repetitive, sure. But it's the same as the Coyote having 15 things fall on his head after he falls off of a cliff.

I think the reason Reddit loves spongebob so much is the majority of users grew up watching it and it's nostalgic.

Cookie clicker flashbacks

I've yet to see any significant real-world uses of AI

Have you ever used Google?

Falling isn't walking, knocking the pieces off the board isn't playing chess.

Hey Reddit what am I now?

Uh stupid?

No I'm pretentious!

What's the difference?!

Hahahahahahhahahahaha

Venerate the immortal Emperor.

/sub/unexpected40k

This clip isn't trying to "prove how stupid Patrick is." It's just showing off his stupidity in a humorous manner.