Are you talking about logic programming?

What I gather is that OP is attempting to describe a system for a learning AI which looks not only at past identical situations, but also at other past situations and what was learned as analogs.

Something like:

I know that an aggressive person is more likely to harm me -> I should avoid aggressive actors; I was once bitten by a wolf and it hurt -> wolves can hurt me; I'm now encountering a dog -> a dog is like a wolf -> the dog is capable of harming me; the dog is postured aggressively -> I know that aggressive actors are cause for caution -> the dog is likely to harm me => ergo, based on past non-identical situations, I can postulate that this dog represents a physical danger to me.

That's an important part of learning to be sure, and one which is some non-trivial A.I. sorcery, however I'm certain the concept is not novel to the field of A.I. either. I don't especially know the proper name for this, or how common/advanced its use is in games (I imagine its uncommon but not rare, and I imagine its limited in scope).

Perhaps someone here more versed in the field can point OP to some research on this technique and OP can compare before describing what the differences are, if any.

In a sense yes.

Any pattern has a set course,drawing a line,raising an arm,running.Logic influence logically modifies the pattern for a specific purpose.

A program has a set course,But,Logic modifications change the course or program.Logic Relative Addressing is just a way to record the Patterns ,or new patterns,

in relative sense so several can be related for data sent to a network.Thus allowing relative logic modifications.

Of course this is an abstract concept,but if it works it works.

My guess is yes.

It sounds to me like you want to apply dynamic logic programming to other areas of software development than AI, specifically as a way of optimizing or specializing arbitrary algorithms to the current situation in which they are running. Is this correct?

Ravyne

Thank you.

You are right.

The difference is that a score is like 10%+3%+7%= ?% of 100% if 70% then pass

This is (part of logica relative to part of logicb relative to part of logicc) and( relative to logicg relative tologicy) is logicn

Now try and explain that.???

Oberon_Command

I'm just an old electronic man that saw something in the parallel resistance formula.

I'm trying to show someone that can understand {1 out of 100) that might find it important information.

I've tried to prove it wrong but with my limited experience have only see things I never saw before.

Didn't want the possibility to die with me.

Thus Onedream to contribute something to make a difference.

At least now it's in your hands.{such as it is}

Thanks for that to you all !!

Sorry if I annoyed any by dropping in.

No worries, its hard to communicate across boundaries of different experience and skillsets.

From the additional details, it sounds like you're wanting to add in weighted factors for the past experiences your AI is considering for the current situation (a past situation that is more like the current situation has more influence on how the AI responds to the current situation), and then there is another factor of "all things considered, is enough about this current situation similar to what I already know, that what I already know should dictate how I respond" which sounds like your basic fuzzy-logic (what confidence threshold do I need to meet before I'm sure of how I should respond). Possibly you also have some feedback loop in there, it wasn't clear.

Its a combination of techniques that I'm sure isn't novel. In my entirely non-expert understanding of AI, it looks to me like a fairly standard neural network, where the trick is that the neural network applied to the situation is built up out of past experiences, and new experiences are added in as you learn.

I suspect that what you've had is one of the Eureka! moments which seem really important to you (and are), but which have already come for people working in the field. I'm sure we've all 'Invented' something we were really pleased with, only to come to find that someone had already thought of it first. Its easy to be sad about that, or even to feel a bit naive afterwards, but I like to take it as a validation personally -- if you came up with something independently and can see that someone else did too, and maybe that's its valuable, then that speaks well of your intellect even if you don't get to take credit.

But again, I'm not expert at all in this field, so if anyone can spot something truly novel here, please say so.

AI that interprets (cognition and factoring) a situation to solve problems is nothing new, and the difficulties for anything more than trivial situational environment (and solution space).are great

Initial learning data has to be provided (raw and bulk) so that it can be factored (filtered/interpretted) for the things which are indicators of the patterns of cause and effect, so that logic can be applied for decisions so that proper actions can be taken to achieve the objects desired goals. The individual Results themselves have to be identified, measured and compared. Just selecting WHICH things ARE the results is a big problem. Similarly deciding which action(s) caused the results has to be done (detecting a linkage patterned between them). Finally based on that factor information and a current situation Choices have to be planned and made and TESTED for success to see if a correct conclusion was made (possibly requiring revision of the causality model that was used). History of all relevant situations must be statistically preprocessed (possibly reevaluated as all run-time happenings get recorded and the learning set is expanded) and the patterns refound/reformulated.

Greater dificulties happen when there is a Temperal aspect - some action (or sequence of actions AND the suituation they were applied to ) in the Past is the REAL cause of current happenings (or pattern of happenings). That significantly increases the data relevant to any current cause and effect pattern.

The difficulty lies in the system NEEDING to first be told what is Good and Bad Result-wise (for the one object acting), and how to quantify those results into metrics so that mathematic logic can be applied for with priorities/probabilities decided. Even in semi-complex simulations, the present situational factors (objects and actions and results) explode combinatorically (they all interrelate to each other, and result in an N^2 expansion, at minimum).

Adding in incomplete or faulty information and the resulting Uncertainty makes doing all the above more than a magnitude more difficult. (just judging whether you HAVE enough information is a separate problem and requires alternate strategies to handle cases of knowing too little).

Add in simulated people, with their own motives/mental states and complicated reasoning/irregularities, and the complexity increases and the AI efforts increase further.

Making this all into a runtime learning process (with limited resources/time) introduces the additional metric of : How much has what I know changed for me to consider reevaluating it and how extensively need that be done - what new data indicating NEW factors and relations requires rebuilding the internal model of how I interpret and predict the simulation 'world'.

The logic for one potential mathematical model for a type of interaction pattern is simply a tool for the process above (many such methods of detecting/defining patterns and modeling ARE needed so they can be applied against the complexity of a non-trivial simulation).

It would help if your spelling was on point. Especially since the incorrect spelling found it's way into your program there. Now while I'm certain I wouldn't have a clue what you're trying to convey either way, it would make me take this a little more seriously. When I see a topic in the game design forum, if the very first word in the title has been so horribly botched as yours is I usually wouldn't even bother to take a look. However you did say you had a new concept, which I always find a hoot as people post all the time thinking they have a radical new idea they just birthed from their loins and it is the highlight of my day to go comment on said post how unoriginal it really is citing all the times I've seen it before and/or what a terrible idea it would prove to be upon implementation into a game. Let's just say I was given no shortage of ranting material with yours.