Quantcast
[ 3 / biz / cgl / ck / diy / fa / g / ic / jp / lit / sci / tg / vr / vt ] [ index / top / reports / report a bug ] [ 4plebs / archived.moe / rbt ]

Due to resource constraints, /g/ and /tg/ will no longer be archived or available. Other archivers continue to archive these boards.Become a Patron!

/tg/ - Traditional Games

Search:


View post   

[ Toggle deleted replies ]
>> No.16248910 [View]
File: 1.11 MB, 1600x1200, 0a6dd0f7508144af4bd48a3ff6f821e191904518.jpg [View same] [iqdb] [saucenao] [google] [report]
16248910

>> No.14479132 [View]
File: 1.11 MB, 1600x1200, Gynoid 16.jpg [View same] [iqdb] [saucenao] [google] [report]
14479132

>> No.13794623 [View]
File: 1.11 MB, 1600x1200, 1286422067055.jpg [View same] [iqdb] [saucenao] [google] [report]
13794623

Fuck off, it totally counts!

>> No.12605765 [View]
File: 1.11 MB, 1600x1200, biohazard girls.jpg [View same] [iqdb] [saucenao] [google] [report]
12605765

>> No.12356197 [View]
File: 1.11 MB, 1600x1200, fed71a3446871d15bdccd8e491426b2a.jpg [View same] [iqdb] [saucenao] [google] [report]
12356197

>>12355994
I'm imagining huge, specialized-use pipelines. Dump your data in seconds flat.

Alternatively, you could have the robot transmit decision state data in realtime. A box running in the sidelines would take the data and run it through an identical AI, thus developing a perfect copy.

The system would be compromised if the droid in combat suffered any damage that resulted in a software hiccup, and probing for bad learned behaviour can be very, very difficult. One solution to this would be to have the parallel box run on a few seconds lag, detect the droid's defeat, and adjust the input accordingly, in a well-THIS-didn't-work-out kinda way.

...I'm currently applying for a master's program to study AI, incidentally.



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