tag:blogger.com,1999:blog-67662364773488817.post8433824827337389442..comments2009-04-23T12:19:21.774-07:00Comments on The Evolutionary Brain Project: The "alien" technology of lifeevobrainhttp://www.blogger.com/profile/00188115394104188830noreply@blogger.comBlogger5125tag:blogger.com,1999:blog-67662364773488817.post-5810292913442483762009-04-23T12:19:00.000-07:002009-04-23T12:19:00.000-07:00As far as I know, there were such attempts, and th...As far as I know, there were such attempts, and the language used was lisp. Indeed, lisp was developed with this goal in mind -- to eliminate any distinctions between data and commands. <br /><br />Nevertheless, I think this idea is worth developing: very complex problems are sometimes solved by people who simply hadn't known that they are complex.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-67662364773488817.post-76390480362998953732009-04-17T12:56:00.000-07:002009-04-17T12:56:00.000-07:00You bring up some very interesting points.
For me...You bring up some very interesting points.<br /><br />For me the biggest problem with the current computer architectures is that the data and the program are completely separate. In a brain there is no distinguishing between the data and the program. The memories determine the future behavior. This is what I am trying to imitate in my "evologic" language. <br /><br />Perhaps we can think of the functioning of a cell in a similar way with the data and the program (DNA) all floating together. I don't know. I would have to give more thought to that. Is DNA the program, or the data, or both?<br /><br />I should also clarify that I think of the learning process as using an evolutionary algorithm. That is why I chose the name "evobrain". The fitness function in this case corresponds to maximizing good feelings while minimizing bad ones. I can get into more details on that if your interested. <br /><br />As I've said, I haven't really worked on this project in about a year. And I really need to review my notes in order to get back into the swing of things.<br /><br />I appreciate your comments. Its helping me to refocus and recapture the original inspiration I had for this project. Fortunately, I did keep very good notes and can pick up where I left off.<br /><br />I don't know about fractal techniques. That's something that would be beyond my abilities at this point. <br /><br />One thing about mutations is that it seems to me that there is a fine balance between too much change from a mutation and too little. Nature seems to have worked out that balance. It seems to me that the majority of mutations in a system should not be so radical as to keep the organism from functioning. That is a goal of mine in creating the evologic language.<br /><br />Normal programming languages are no good at mutating at all. Just a semi-colon out of place will cause a program to cease to function. I'm trying to create something more like DNA. Perhaps this is more like what you would call a "system" than a "language". But I'm applying that DNA idea to memories. Then the memories become the program and evolve over time based on experiences (new memories).<br /><br />That's the basic idea anyway and that is how I am hoping to create an artificial intelligence that can learn and adapt based on its experiences. This is my understanding of what the ant does and this is what I am trying to emulate.Frank Hopehttps://www.blogger.com/profile/01610256051568976952noreply@blogger.comtag:blogger.com,1999:blog-67662364773488817.post-64508607292622422009-04-17T03:07:00.000-07:002009-04-17T03:07:00.000-07:00The idea of evolution is of course interesting, bu...The idea of evolution is of course interesting, but for me what really distinguishes life from technology is individual's ability to adapt its behavior and learn from its experience. An ant might be using not so sophisticated receptors, but it uses the data very effectively, and that's with so little brains!<br /><br />I think it is the ability to learn and use the 'actuators' flexibly that is really needed for the technology to proceed to the next level.<br /><br />Nevertheless, we need evolutionary mechanisms for long-term self-improvement of machines. And I agree that choice of the language to implement an evolving system is important. But I think it is not the 'language', but rather entire 'runtime environment' that needs to be developed.<br /><br />The problem is that in almost all languages programs are executed in sequence. Each instruction has strictly defined behavior. It isn't the 'natural way'. There we have a DNA molecule and lots of stuff floating around. The 'execution' is governed by very complex processes (i.e. electrostatic forces), and the program gets executed as a whole, not in sequence. This can also account for the amazing data compression (as I said 10 MB is enough for almost every organism).<br /><br />You probably should try some fractal techniques for execution of programs: in fractal graphics, programs are usually very small yet yield very complex results, and single mutations can significantly change the behavior.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-67662364773488817.post-53790136317432245002009-04-15T20:46:00.000-07:002009-04-15T20:46:00.000-07:00I don't know how many megabytes of software is in ...I don't know how many megabytes of software is in a elephant, but I can tell you that the end result is much more impressive than any human software that exists today. The fact that human generated code is difficult to understand does not necessarily make it interesting or worth understanding. <br /><br />What makes biological life interesting and unique is its ability to adapt to its environment. We see this through life's ability to heal from wounds, to learn, to evolve. Life doesn't just replicate. Each individual is unique, yet each individual is able to function fully.<br /><br />I have been considering the possibility recently that part of what makes life unique is the "language" in which it is written. In this case I'm using "language" in the sense of a computer "language". I have had in mind for quite a while to write an evolvable language which I call evologic. That's what I'm planning on working on in the near future.evobrainhttps://www.blogger.com/profile/00188115394104188830noreply@blogger.comtag:blogger.com,1999:blog-67662364773488817.post-67013162112010170942009-04-07T02:58:00.000-07:002009-04-07T02:58:00.000-07:00There's really little reason not to call an elepha...There's really little reason not to call an elephant 'machine'. Furthermore, isn't an elephant a piece of software with 10 megabytes of source code? (http://genomicron.blogspot.com/2007/11/bacterial-genomes-and-evolution.html)<BR/>People create software of that size, so if the life was 'coded', wouldn't it be equally difficult for the Developer to understand our software?Anonymousnoreply@blogger.com