HentHighSchool Development Forum

Please login or register.

Login with username, password and session length
Advanced search  

News:

collapse
* Recent Posts
Re: [HHS+ 1.9.5] Official Bug Report Thread by Shilo
[Yesterday at 07:42:57 pm]


Re: [Mod] King's Way 2.0 by Spellsword
[Yesterday at 05:21:58 pm]


Re: [Mod] King's Way 2.0 by vadi92
[Yesterday at 04:13:53 pm]


Re: [Mod] Grafitti and interrogation (function) by fateavernicus
[Yesterday at 02:05:31 pm]


Finished the events/game/entire town enslaved by you! by fateavernicus
[Yesterday at 02:00:29 pm]


Pages: [1]   Go Down

Author Topic: Neural networks in corruption games  (Read 9224 times)

[xox]

  • Full Member
  • ***
  • Offline Offline
  • Posts: 221
    • View Profile
Neural networks in corruption games
« on: February 25, 2016, 04:28:31 am »

Basically, I'm thinking that by using neural networks corruption games could be much more interesting as they could result in something like various personalities. Another benefit would be, that you could actually properly train the actors. Actors could also have fetishes of various degrees that you could amplify or suppress.

A huge disadvantage is the amount of process power needed. Also the learning seems to be very slow. Not sure if this completely disqualifies the use of NN in games.

Are Neural Networks possible to be used? Did someone try it? Does anyone have generally experience with neural networks?

https://en.wikipedia.org/wiki/Artificial_neural_network
Logged

WarOwl280

  • Newbie
  • *
  • Offline Offline
  • Posts: 41
    • View Profile
Re: Neural networks in corruption games
« Reply #1 on: February 25, 2016, 05:11:42 am »

if you could thread the programming you could use some of the multi core processors to get some of it . Nothing like true AI . But at the end of the day most games are not Real AI but a set of programmed responses to certain set of actions . Cheaper programming wise would be to use players for both sides of the coins . Which gets you to wetware verse hardware question . If you want 1-200 responses to certain actions you really dont need AI or NN .

 
Logged

sivael

  • Guest
Re: Neural networks in corruption games
« Reply #2 on: February 25, 2016, 12:56:40 pm »

Neural networks unfortunately have to operate on very large samples to get decent results.
If there's a 1000 events daily that push a neural network in direction A (Say society doesn't like nudity) and 1 that (albeit strongly) pushes the character the other way, the 1000 events will teach the NN to do as society wants it to.

Lacking the broad spectrum of push/pull affectors real people have (morality, religion, interpersonal relationships, memory, history, will, personality etc...) it would just make the whole society homogenize.

To make it interesting for the player we either cheat (make 10 stats per person and modify them) or we do a full AI and that's not happening.
Logged

Shilo

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2217
    • View Profile
Re: Neural networks in corruption games
« Reply #3 on: February 25, 2016, 05:48:17 pm »

I've created a multithreaded modular neural network before, for stuff like function approximation in machine learning, i.e. getting plenty of feature values from plenty of samples and training the network to give me a single scalar result for any given sample. I just don't see how that would benefit the game in any way.
Logged

Tomas1221

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 62
    • View Profile
Re: Neural networks in corruption games
« Reply #4 on: December 26, 2019, 03:55:41 am »

it sounds fun but too complicated overkill. ant then you think about processing power needed. but i had different idea if would be possible to make hhs 3d edition and make emulator that rewrites hhs event to 3d engine. taking basic like from Artifical academy for visuals but reading event scripts from hhs
Logged

DASh

  • Newbie
  • *
  • Offline Offline
  • Posts: 4
    • View Profile
Re: Neural networks in corruption games
« Reply #5 on: February 20, 2020, 09:39:08 pm »

I've been playing around with basic AI algorithms for some years, myself, and I could see several ways to actually use machine learning the way you're thinking, xox. The problem, as you said yourself, is processing power, but even that could be somewhat reduced by having each character controlled by their own (small) "neural network" instance - the quote marks being there because I'd probably use something more primitive in this case. With a couple hundred characters in an average game, though, it could still take a minute or two between turns...

... unless we're talking about a multiplayer game hosted on a sufficiently powerful server, or distributing the character action calculations among the connected clients. But even if it were based on the same events etc., we'd be talking about a completely different game here, basically. Probably one where each player is a student instead of the principal... Or maybe a parent. And it might weird people out that they wouldn't know whether a specific other character in an erotic game is played by a human or a computer. Might still be fun, both to play and to create, but to be honest, I lack the motivation to do so in what little free time I have. And I get stumped on finding a good way to handle players leaving the game. And yes, that means I've been thinking about this for a while before now.
Logged
Pages: [1]   Go Up
 


anything
SimplePortal 2.3.3 © 2008-2010, SimplePortal