Archived

This topic is now archived and is closed to further replies.

Cadet Random

Needs vs Schedules

Recommended Posts

Cadet Random    122
(PS [that''s prescript today, not postscript ] Hi guys, I''m Sam, new here, i''ve posted a couple of posts w/o a registration, but be seeing more of you, probably) A scheduling system can be indirectly derived from a Needs based system. Ie: Need: Food [+continual need] Action: Buy food from shop Action: Eat food Subneed: Money [+continual need inherited] Need: Money [+continual need] Action: Work Subneed: Job [+continual need inherited] Need: Job [+continual need] Action: "Get" Sub-Action: (Whatever, it''s keep that''s important to my point) "Keep" Sub-Action: Turn up to work on time every day Thus, turning up to work on time will become part of the need dervied actions, and work (or school) for most of us forms the basis of our daily schedule. For instance lunch Action: Be at work on time at the right times: 9-12:30, 1-5:30, monday to friday. Priority: 1 Action: Eat food Priority: 2 And thus, somehow the mechanism would fit eat food into 12:30 til 1, and dinner after 5:30 Additionally I think the need -> action rules should all be learned by the system. Writing these would be tireseome, and it''s emergent behaviour we''re looking for right? Plus this would reflect a "wisdom of age" part automatically. There are plenty of machine learning algorithms around to choose from. Of course, at the beginning they''re all going to be bloody stupid and starve to death pretty quickly. So, you add a mechanism that when two agents spend time together, their rules are shared. Combined with reproduction (which''d be easy as, need for sex and random pregnancies would be the "nicest"), this would do the "learn from parents" type thing. Tho, not directly. Somehow you want the default behaviour to be one that initally is set up as a "parents are god" type scenario, which gradually decays (or gradually decays, drops off totally after 13 years for 3-5, and slopes back up to a midpoint - modelling adolescence But, then, shock horror! this is predefined action! We could use Genetic Algorithms to control that piece of the puzzle Hehe, this is starting to sound like fun fun fun As always, your thoughts? Sam Space Cadet''s DJ Random, Thinker, and Geek at heart.

Share this post


Link to post
Share on other sites