[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/vt/ - Virtual Youtubers

Search:


View post   

>> No.60181948 [View]
File: 299 KB, 700x516, 1694806970287778.png [View same] [iqdb] [saucenao] [google]
60181948

>>60180890
Yep. The "books" help the AI to estimate when it doesn't know the answer to a problem. The more books it got given, the better it is at estimating.

The biggest difference between classic programs and neural networks, is the classic programs follow a determined flow. For example:
If ("car is red") then say: "the car is red!"
else, if ("car is blue") then say: "the car is blue!".
So when it gets a green car, it doesn't know what to do.

Meanwhile the NN gets a mixed bunch of car pictures and learns to estimate what color a car is based on what it learnt. So it can guess a green car. The more "picture books" you fed it, the better it is at guessing car colors even if it never saw them before.

Navigation
View posts[+24][+48][+96]