WeirdoFu

Members
  • Content count

    362
  • Joined

  • Last visited

Community Reputation

205 Neutral

About WeirdoFu

  • Rank
    Member
  1. I just can't resist the urge.....I have to say something. There really is no argument when it comes to defining terms like crossover, mutation, selection, etc. because they have all been defined in literature and they all fall under the umbrella of evolutionary computation. There is also an agreed upon definition of an evolutionary algorithm (EA) out there (dating back to 1993). Just to point out a few things. A GA without crossover is not a GA. It is actually an evolution strategy (ES). And the cool thing about ES is that it varies the amount of mutation based on successes of previously applied mutations. However, later on, crossover was added into ES in one form or another to increase its robustness. Other additions to ES include complex covariance matrices to properly guide search, which at that point, I find it pretty much overkill. Now as to the merits of crossover, that is arguable, but for all EAs to work properly, it all boils down to exploitation vs exploration. The point of crossover is to promote exploration of the search space by incorporating partial solutions from other candidate solutions. Mutation is simply a form of exploitation of the current solution and local solution space. However, depending on the convergence of the population, the roles of the two can flip around. So, going back to a mutation only EA. To make it work well, you must somehow counterbalance the exploitative nature of mutation with some other mechanism that promotes exploration. In the case of the original ES, it varied the amount of mutation to try and promote exploration, but it seems that comparably, crossover created more exploration, which is why few stuck with the original ES structure. Back to the original posters question about genetic programming. As cool as that would be to have GP in games, chances are slim that it would become practical. Though there may be some use for it somewhere in the development pipeline. Also, note that GP does not have to evolve actual pieces of code. Most practical applications of GP out there involve using it to do curve fitting by evolving the actual mathematical equation for the curve or data in question. More useful for game AI would probably be Evolutionary Programming, which, if memory serves me correctly, was originally designed to evolve finite state machines, which has a much higher level of relevance when it comes to practical game AI. So, for example, evolving the state machine that defines an AI agent's behavior on the fly through gameplay feedback. In the past decade or so, people have been trying to improve on EAs by pretty much mixing and matching everything they can come across, so no one really works exclusively with any of the 4 originals anymore (GA, ES, EP, GP). At this point, the lines are so blurred that it is quite pointless in arguing whether something is a GA or not. The fact is that they're all evolutionary algorithms in one shape or form. Oh yeah, and for people who are obsessed about gradient ascent/descent, there is the Society of Hill-Climbers and the evolutionary society of hill-climbers that is definitely worth checking out.
  2. Interestingly enough, the upcoming game, Brutal Legends, will have a similar RPG-RTS hybrid as a multiplayer mode, where you command small squads of units, while you yourself can also jump into the action.
  3. Quote:Original post by Durakken "So i have never played any star ocean game before and i just got this one a couple weeks ago. i have played it once or twice but i seem to be stuck where to go.. i just started the game and i was told to find the other ships that crashed but the only place i can seem to find is urd falls cave and it seems like everything else is a dead end...i must say i dont feel like the smartest person with this but i just want to get past the point so i can actually enjoy what looks like a really fun game. any help? " For those of you have played the game... you can only go from your ship to the waterfalls, through the falls, and to another part of the map to get where you need to go. The direction you should go as well as the entrance/exit is clearly marked. It doesn't take a genius to figure out that if it's the only available path to follow then it should be the path you take... This isn't the only example of this I've seen. Playing MMOs, especially ones like CoH, the developers go out of there way to have tutorials and way points and markers all over the place so you don't have to ask someone else, but I find people that without exploring their HUD or anything around shout they're lost and are confused and want another player explain to them what was in the tutorial provided by developers... I've always found this behavior incredibly disgusting and been annoyed with it. Am I the only one that feels this way? I ask this here because it's a design element that costs time that for many just is a waste...so what do you all think about this behavior? To answer the original question, since I don't seem to see a response from anyone who have actually played the game, the key is on the mini-map, which you can make full screen at anytime by pressing the start button. The mini-map is usually up in the upper right hand corner and will actually display context sensitive details about the environment. For example, in your case, go where there is a big exclamation point. As someone who has played the game for more than 60+ hours, I think their choice to move the important information about the environment from the actual displayed environment is to preserve the sense of realism. You have a game where you are usually running around wilderness settings on undeveloped alien planets more than half the time. So, it makes sense that you wouldn't see crazy markers in the world that resemble platforms or road signs. They even replaced the concept of a "chest" based on the environment around you. So, in an insect infested area, a "chest" may be a egg pod like thing. The only thing the game is guilty of, along with some of the recent entries from tri-Ace is having itty-bitty print and on screen text that you could not possibly read on a standard def TV. This also is true for the markers on the map, they aren't that visible on a standard def TV, especially if it is less than 30". Move to HD and the game looks and feels totally different. So, I know the arguments here have skewed towards a debate over level design, but I kind of wanted to address the original question and say that it was most likely a design choice to not pollute the environment with artificial markings. It should be noted that this issue mainly occurs in the wilderness areas of the game and is less so in the indoor artificial environments, like in a space station. So, I guess the simple bottom line is, how much of the immersion do you want to break in order to point the user in the right direction when it comes to level design. How artificial can you make an area before it feels forced and out of place with the rest of the environment and proposed setting.
  4. Good AI equals self-awareness

    Quote:Original post by polyfrag Quote:Take pain, for instance. Pain is something that we experience frequently when we are younger. Why? Because our brains have not yet learned to avoid things that are harmful, such as hot objects. Experiencing pain is, as we all know, very unpleasant; it provides feedback to the brain saying "that was bad, don't do it again". But free will also comes into it. Lying in bed I may feel an itch on my cheek or may feel slightly uncomfortable in the position I'm lying. How does the mind create the qualia of "not liking" this? I have a choice of whether to scratch my cheek or not and whether to try to find a more comfortable position. But how does my nature incline me towards doing this? It seems that these things are like the "natural" things to do and my choosing to resist scratching the itch or moving to a more comfortable position is action that is "against the current". I think maybe that as children we feel these qualia that give us an ambiguous, nebulous sense that requires some action that something isn't right that won't leave our attention and won't let us focus on other things unless we attend to the things causing our state. And through actions eventually we learn over time to how to "relieve" these unpleasant states. "Relieving" an "uncomfortable state" is not free will, It is the decision to do the opposite that is free will because it is an active decision to go against a reflex. The whole concept of "not liking" something is actually a conscious categorization of unconscious or subconscious behavior or reflexes. There are quite a few hardwired behaviors of the brain and the physical body that comes from evolution. Take your example of an itch. At some point in evolution, it probably became a survival necessity to be able to feel light contact on the skin, which induced what we call an itch. The natural response is to get rid of it. Humans aren't the only animals that feel the need to scratch an itch. Most all mammal have the same behavior. The point of the skin and its ability to feel pressure is to warn us of environmental hazards. All reflex responses are fundamentally built on the need for survival. So, on the lowest level, there is pain, hunger, etc, which directly affects survivability. Then built on top of that, once survival has been taken care of and guaranteed, there is then the "quality of survival" or "quality of life" layer. And it is on this "quality of life" layer that most of our reflexive emotional responses are built on top of. So, the purpose of behaviors on this level is to further maximize survivability by creating behaviors that decrease the chance of having to trigger lower level reflexes. As the layers grow, their dependencies on previous layers become more complex, which is why it becomes less apparent how certain behaviors can be attributed to survival instincts. So, ironically, true "free will" is the conscious choice of going against our instincts. However, there are times, where your seemingly conscious decision to go against a certain instinct or reflex is because of some other higher level behavior having slightly greater benefits when performed. So, for example, instead of scratching the itch, you get up and go take a shower instead because you realize that may solve the root problem. So, this further begs the question of what are choices made of "free will" and what are choices that are based on established behavior patterns from experience.
  5. Ideas for a Modular Enemy AI

    Two Word: Subsumption architecture http://en.wikipedia.org/wiki/Subsumption_architecture
  6. Good AI equals self-awareness

    For once I would like people to stop trying to anthropomorphasize everything and just dump the idea of making machines "human like." Though that won't be possible since humans are ego-centric beings that once thought the universe revolved around them. Well, I guess we haven't stopped thinking that either.
  7. Your best bet is to try a few different methods to see which gives you better results. The method you've described should work. Just make sure you balance exploration and exploitation such that you don't get stuck in a local optimal for too long. Though, in most cases, local optimal are unavoidable. One things to note is that you may have to play with your evaluation criteria. You may not just want winning the match be the final deciding factor. Down the line, you may need functionality to do a point-wise evaluation system. So, for example, winning the match may give a candidate alot of points, but being able to destroy just about every enemy unit may also be a valuable trait. This way, you open up the process to being able to generate the "optimal" win. It also gives you the ability to evolve different AI's with differing goals. One may want to try to beat the opponent with a rush, while another will just play defensive all the way through. Just remember, release the game with the values and not the EA system. One interesting idea to try is to have 2 tournaments. One is to evolve the best strategy to win, while the other tries to evolve strategies to beat those from the first tournament. This way, you're not just trying to find a good strategy, but a robust one that can stand up against different types of attacks.
  8. Quote:Original post by PlayStationX 1st: any system with 'B', i.e. memory(input), is flexible system, it has evolutionary potential Memory does not make a system more flexible, especially if that system is a human being. Memory is the basis for assumption building, and decisions are made based on past memories. Memories shape and alter our perception of the world. So, I find it hard to agree that memory facilitates a flexible system. It seems to be quite the opposite. You forget that humans are not rational beings and that the human mind is great as finding a path from point A to point B even if it means the path passes through preconceived notions, assumptions about the problem, personal biases, or even external influence. Memories define who we are while binding us at the same time. Quote: The first point is to show how "life" and mental processes are not something to be measured in boolean with yes/no, but rather with floating point or percentage. I think most AI people agree with this by now, even most common people. Quote: The second point is to define all of the following in the terms of information technology and above virtual definition using concepts of information: input<->memory<->output->reaction: cognition consciousness perception reflex detection emotion feeling thought instinct anticipation decision acknowledgment knowledge willingness comprehension intelligence awareness intuition recognition intention sensation mind self-awareness Sounds like a definition for an intelligent agent. Quote: The third point is about Turnig machines, determinism, quantum uncertainty, algorithmic predictability, "free will" or "active will", possibility of simulation of self-awareness and about practical means to measure it. The bonus point is to define and try to answer the concepts of anticipation, intuition, prediction and prophecy. Ah, the eternal argument of free will. Personally, I believe there was never free will in the first place, just the illusion that there is and was. The outcome of a coin flip can be determined if all variables involved are known. The outcome of "flipping" a human can also be predicted if all variables involved are known. It is pretty much the basis for most magic tricks and mind tricks.
  9. Character development culture

    If you want to incorporate it into the culture, then there are a few aspects to consider and throw around. Firstly, if it is within the means of society to measure ability, then you will have a counter-culture somewhere where they reject these ratings. These counter-movements will naturally be made up of people who have very low "stats" at birth and were disown, exiled, or just forsaken by society and of people who are sympathetic to these outcasts or just plain want to be different. There will also be religions establishments around some warped version of the mainstream based on some weird, supposed, hidden "stat" (scientology?). There will also need to be a black market for artificial stat enhancements. On top of that, you can build a government conspiracy to build machines that extract certain "stats" from people and give it to others for the sake of creating a super-human or army of super humans. So, yeah, you can wrap everything you see in modern society into this, since we all know that human nature never changes. Here is one thing that can spice things up. The inclusion of a known, yet unmeasurable "stat(s)". Things that cannot have quantitative values attached to them and seemingly have no visible effect. They are visible stats with no known measurable value and with seemingly chaotic and random hidden effects. It can be the big question mark that can be used to drive the story and even the game.
  10. Dynamic Vendor Pricing

    Both Suikoden 3 and 4 have a similar system of commodity trading in game. The player is encouraged to buy raw material in one region and sell it in another that needs it more for a higher price. The prices fluctuate due to supply and demand, so you can actually flood an area with an item and cause it to drop in value. It is one of the main ways of making money in the game. And just like stocks, if you hold on to something for too long, it may end up becoming worthless.
  11. Difficult game controls

    Try Robot Alchemic Drive (RAD) for PS2. I always use that as an example of a fine tuned control scheme with a steep learning curve. In the game, you essentially control a giant robot in battle from a 3rd person perspective. Controls are as follow: R1/R2 - move right foot forward and back L1/L2 - move left foot forward and back Left Analog Stick - swing/control left arm Right Analog Stick - swing/control right arm Direction Button - rotate torso face button - toggle weapons So, to walk, you literally have to move your left and right foot forward one after another. To turn you move one foot forward and one foot back. To throw a punch, you pulled back on a analog stick, then push forward. No fancy key combos, just alot of minute controls.
  12. What could you do with a 'many-core' CPU

    http://www.sandia.gov/news/resources/releases/2009/multicore.html
  13. are certain Clichs good?

    The scary thing is, the more I read about what he picked up on, the more your story sounds like Kingdom Hearts. "The main character living on a tropical island" Check. "Getting swallowed by darkness (the same thing happens to your best friend), getting transported from that island to another world, meet two people who agree to help you look for your friend if you help them look for their older brother" Sora and Riku were transported to another world where Sora meets Donald and Goofy who are looking for Mickey. The relationships kind of work out too. So, check again. "Having the main character being the chosen wielder of a special sword type weapon " Sora is the wielder of the Keyblade. So, check again. "Having the settings be different worlds(not planets, worlds that are completely separate)" They travel to different worlds based around the many Disney movies/stories. Check. "The main stabbing his heart, with a special blade. fragmenting his heart(metaphorical heart not the actual organ), causing himself to turn into a mindless creature, all to awaken his best friend,whose heart has become part of the main characters (again metaphorical) who is a girl, who he secretly has romantic feelings for(This would make ALOT more sense if you read the story, PM me of you want of cop of the story document)" There is a portion of the story where Sora becomes a Heartless, which matches your description pretty well as well. "The main character being brought back to his human form by the friend:who is a girl(she hugs a random creature who she thinks is the main character, a flash of light and he`s back to his human form)" Forgot how Sora got brought back, so iffy if this matches. "Having the two main characters be the key to opening a door to the world were all of the remnants of other worlds are that houses a tremendous source of power at the end of it" The door way to Kingdom Hearts is kind of similar to what you're talking about. Not going to really say it is a match there. "Having a villain who wants to get said source of power." This is a given for all games though, especially one that involves some sort of ultimate power. There are cliches and clones, and from the stuff you've listed, your idea seems closer to a clone.
  14. Games with "Unique" PoVs

    Quote:Original post by Zouflain lol. I just watched Eagle Eye, and while I found the movie forgettable, I think that's a unique perspective that games rarely (if ever) utilize. Rather than having a fixed perspective, instead you could have a game where you hijack/hack into perspectives, or manipulate the environment to create new perspectives (going off of the inspiration, say, coercing someone to move a camera to a new position). That would be interesting. A resident Evil 1 like game where you are actually remote controlling a robot/person from a secure point. And depending on which security cameras you gain access to, you'll get different perspectives to work with and in turn create your own blind spots.
  15. ga initial population

    Interestingly enough, what you are looking for seems to be similar to a steiner triple system from combinatorial design. There is an algorithm that can be used to generate the solution set (a set of triples) along with a way to determine exactly how many triples you'll have. I'll need to look it up to get the details, but that seems to be what you want to look for. Also, the standard generating technique uses a idempotent latin square.