[agi] Tired not making any MONEY online ! 6286SmBA5-781ZTm-15
Dear fellow marketer, You get emails every day, offering to show you how to make money. Most of these emails are from people who are NOT making any money. If you want to make money with your computer, then you should hook up with a group that is actually DOING it. We are making a large, continuing income every month. What's more - we will show YOU how to do the same thing. This business is done completely by internet and email, and you can even join for free to check it out first. If you can send an email, you can do this. No special "skills" are required. How much are we making? Anywhere from $1000 to $9000 per month. We are real people, and most of us work at this business part-time. But keep in mind, we do WORK at it - I am not going to insult your intelligence by saying you can sign up, do no work, and rake in the cash. That kind of job does not exist. But if you are willing to put in 10-15 hours per week, this might be just the thing you are looking for. This is not income that is determined by luck, or work that is done FOR you - it is all based on your effort. But, as I said, there are no special skills required. And this income is RESIDUAL - meaning that it continues each month (and it tends to increase each month also and it's LEGAL too!). Interested? I invite you to find out more. You can get in as a free member, at no cost, and no obligation to continue if you decide it is not for you. We are just looking for people who still have that "burning desire" to find an opportunity that will reward them incredibly well, if they work at it. To become a free member and learn more about our unique internet opportunity, simply reply to: [EMAIL PROTECTED] and put in subject: "SIGN ME UP, REF: 123" I will then get back to you asap with further instructions. Looking forward to hearing from you! Sincerely, JAB P.S. After having several negative experiences with network marketing companies I had pretty much given up on them. This is different - there is value, integrity, and a REAL opportunity to have your own home-based business... and finally make real money on the internet. Don't pass this up..you can sign up and test-drive the program for FREE. All you need to do is become a free member! = = = = = = = = = = = = = = = = = = = = = = = = = = = Unsubscribing: Send a blank email to: [EMAIL PROTECTED] with "Remove" in the subject line. This message is not intended for residents of the state of Washington, and screening of addresses has been done to the best of our technical ability. If you are Washington resident or otherwise wish to be removed from this list, just follow the removal instructions above. 4705BYUM2-779nAvY3767aVJa2-526Gnzd7933SmWw3-32l43 --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/?[EMAIL PROTECTED]
Re: [agi] Diminished impact of Moore's Law on AGI due to otherbottlenecks
On 1/4/03 3:02 PM, "Shane Legg" <[EMAIL PROTECTED]> wrote: > > I had similar thoughts, but when I did some tests on the webmind code > a few years back I was a little surprised to find that floating point > was about as fast as integer math for our application. This seemed to > happen because where you could do some calculation quite directly with > a few floating point operations, you would need more to achieve the same > result with integer math due to extra normalisation operations etc. I can see this in some cases, but for us the number of instructions is literally the same; the data fields in question could swap out floats with ints (with a simple typedef change) with no consequences. We do have a normalization function, but since that effectively prunes things we'd use it whether it was floating point or integer, and it is only very rarely triggered anyway. I guess the key point is that we aren't really "faking" floating point with integers. It is a case of floating point bringing nothing to the table while offering somewhat inferior performance under certain conditions. The nice thing about integers is that performance is portable. I certainly wouldn't shy away from using floating point if it made sense. It is just a mild curiosity that when all is said and done, nothing in the core engine requires floating point computation. > I was also surprised to discover that the CPU did double precision > floating point math at about the same speed as single precision floating > point math. I guess it's because a lot of floating point operations are > internally highly parallel and so extra precision don't make much speed > difference? I believe this is because current FP pipelines are double precision all the way through generally. If you run single precision code, it uses up just as many execution pipelines as double precision. The exception is the SIMD floating point engines (aka "multimedia extensions") that a lot of processors support today. But I normally just write all floating point for standard double precision execution these days. > Anyway, the thing that really did affect performance was the data size > of the numbers being used (whether short, int, long, float, double etc.) > Because we had quite a few RAM cache misses, using a smaller data type > effectively meant that we could have twice as many values in cache at > the same time and each cache miss would bring twice as many new values > into the cache. So it was really the memory bandwidth required by the > size of the data types we were using that slowed things down, not the > time the CPU took to do a calculation on a double precision floating > point number compared to say an a simple int. A good point, and one that applies to using LP64 types as well. The entirety of our code fits in cache, but data fetches are unavoidably expensive. > I'd always had a bias against using floating point numbers ever > since I used to write code 15 years ago when the CPU's I used > weren't designed for it and it really slowed things down badly. > It's a bit different now however with really fact floating point > cores in CPUs. One consideration that HAS gone into maintaining a pure integer code base is that it can run with extreme efficiency as currently designed on simple integer MasPar. When used like this, the opportunity exists for scalability that is far beyond what we could get if we required a floating point pipeline. The idea of having scads of simple integer cores connected to a small amount of fast memory and low latency messaging interconnects is appealing and our code is very well suited for this type of architecture. Fortunately there seems to be companies starting to produce these types of chips. Ultimately, we'd like to move the code to something like this, and since there is no design or performance cost to only using integers on standard PCs (they work better anyway in our case), we haven't introduced floating point into the kernel without a good reason. So far we haven't actually come across a need for floating point computation in the kernel, so we've never had to deal with this issue. Cheers, -James Rogers [EMAIL PROTECTED] --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/?[EMAIL PROTECTED]
Re: [agi] Diminished impact of Moore's Law on AGI due to other bottlenecks
James Rogers wrote: We just don't need that much raw CPU to get the job done. The entirety of our number crunching is integer domain, and just addition/subtraction at that. Aside: Since the engine operates on relative values and the dynamic range of integers on your average machine is quite high (higher than anything going on in the human brain), there is no reason to NOT use integer formats, particularly since integer math and manipulation is very fast on most common hardware. I had similar thoughts, but when I did some tests on the webmind code a few years back I was a little surprised to find that floating point was about as fast as integer math for our application. This seemed to happen because where you could do some calculation quite directly with a few floating point operations, you would need more to achieve the same result with integer math due to extra normalisation operations etc. I was also surprised to discover that the CPU did double precision floating point math at about the same speed as single precision floating point math. I guess it's because a lot of floating point operations are internally highly parallel and so extra precision don't make much speed difference? Anyway, the thing that really did affect performance was the data size of the numbers being used (whether short, int, long, float, double etc.) Because we had quite a few RAM cache misses, using a smaller data type effectively meant that we could have twice as many values in cache at the same time and each cache miss would bring twice as many new values into the cache. So it was really the memory bandwidth required by the size of the data types we were using that slowed things down, not the time the CPU took to do a calculation on a double precision floating point number compared to say an a simple int. I'd always had a bias against using floating point numbers ever since I used to write code 15 years ago when the CPU's I used weren't designed for it and it really slowed things down badly. It's a bit different now however with really fact floating point cores in CPUs. Cheers Shane --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/?[EMAIL PROTECTED]
Re: [agi] Diminished impact of Moore's Law on AGI due to otherbottlenecks
On 1/3/03 11:12 PM, "Gary Miller" <[EMAIL PROTECTED]> wrote: > If these benchmarks were more readily available it would be even more > apparent to businesses and users that a 3Ghz machine will not process a > typical application twice as fast as a 1.5Ghz machine. If you actually look critically at the various system performance parameters, you find that system performance these days is scaling almost perfectly with memory performance. While memory performance and processor speed track each other only very roughly, you find that system performance on many benchmarks and memory performance track almost perfectly. There is a strong implication that for many types of applications these days, memory performance IS the system performance "rate limiting factor". The only reason faster processors seem faster is that GHz upgrades frequently come with memory performance upgrades as well. This is also why people who do scientific computing often use the STREAM benchmarks (which measure system memory performance) to compare systems when building a supercomputing cluster. STREAM metrics map more closely to real world performance than running a tight code loop in the cache to measure CPU performance. > How many of you out there with AGI projects feel you are limited > currently in your research by CPU speeds. I was myself up until 2 years > ago. If you do feel you are limited, what speeds are you currently > running at and how much more CPU (2x, 4x, 8x, ...) do you feel you could > optimally utilize. We don't feel particularly CPU limited, but we do feel memory limited, both in terms of how much we can have and how fast the CPU can access it. In fact, it would almost be more acceptable performance-wise to have a cluster of relatively slow processors that can access their (much smaller) memory spaces very fast than to have blinding fast processors with huge address spaces that may take a second or more to traverse in practice. We just don't need that much raw CPU to get the job done. The entirety of our number crunching is integer domain, and just addition/subtraction at that. Aside: Since the engine operates on relative values and the dynamic range of integers on your average machine is quite high (higher than anything going on in the human brain), there is no reason to NOT use integer formats, particularly since integer math and manipulation is very fast on most common hardware. You don't need to materialize floating point values from relative integer values to get perfectly correct statistical behaviors, a point that most people overlook (or perhaps their particular design architecture forces them to do things this way -- hell if I know). But yeah, give me fast memory access and lots of it, and I'm happy. Ironically, most systems made today that really do have good memory architectures also tend to be highly optimized for floating point performance at the expense of integer performance. > And how much do you feel this would compress the actual project plan for > your project. These numbers should give a better indication on whether > and how much current processing speeds are limiting the quest for AGI. The problem today is almost entirely a software problem. The hardware problems and limitations, such as they are, are resolving themselves just fine. The software on the other hand... -James Rogers [EMAIL PROTECTED] --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/?[EMAIL PROTECTED]
RE: [agi] Chess Master Theory Of AGI.
This will occur before the predictions of the experts in the field of Singularity prediction because their predictions are based on a constant Moore's Law and they over estimate the computational capacity required for human level AGI. Their dates vary from 2016 to 2030 depending on whether they are using the 18 month figure or the 12 month figure. Moore's Law is currently at 9 months and falling. My calculations based on a falling Moore's Law put the Singularity on April 28th, 2005. This human level AGI in a computer will be quite superior to a human because of several advantages that machines have over gray matter. These advantages are: upgradability, self-improvement through redesign, self editability, reliability, functional parallelism, accuracy, and speed. This superiority will be quantitative not qualitative. It will be superior but completely comprehensible to us. The belief in a radically different form of advanced thought incomprehensible to present humans is philosophical in nature, not based on evidence. Mike, Is it really true that Moore's Law is at 9 months and falling? Do you have some references on this? Even if this were the case, it wouldn't cause the Singularity by 2005. Processing power is not the only bottleneck! It's true that with faster, cheaper processing power, more people will be able to experiment with more significant AGI systems. But even with a correct AGI design, and adequate funding, computing power and staffing, I think it's going to take anyone several years to get from AGI-design to teachable human-level system. That is the nature of engineering complex software systems based on complex ideas. And of course it may take some time to get from teachable-human-level system to superhuman-level system as well !!! ;-p So, I think that the most wildly optimistic projection we can rationally hope for is superhuman intelligence (the "Singularity") by 2010. But this could only be achieved if *everything goes right* And of course, I don't know how to estimate the odds that everything goes right. An example of "everything going right" would be: One of the currently in-development AGI designs (say, Novamente or A2I2 or NARS) turns out to be almost entirely correct, AND, gets adequately funded... and, teaching a human-level AGI to productively self-modify toward unlimited intelligence turns out to be a matter of a couple years, not a decade. This is a lot of ANDs, Mike -- an awful lot of ANDs ... -- Ben
RE: [agi] Diminished impact of Moore's Law on AGI due to other bottlenecks
> How many of you out there with AGI projects feel you are limited > currently in your research by CPU speeds. I was myself up until 2 years > ago. If you do feel you are limited, what speeds are you currently > running at and how much more CPU (2x, 4x, 8x, ...) do you feel you could > optimally utilize. > And how much do you feel this would compress the actual project plan for > your project. These numbers should give a better indication on whether > and how much current processing speeds are limiting the quest for AGI. We are limited tremendously by CPU speed and RAM capacity. Either greater CPU speed or greater RAM capacity would be valuable, but the biggest boost would be both together. We could utilize essentially any amount of CPU speed or RAM capacity. No limit in sight. Having a CPU with (for example) 10x greater speed would have a HUGE positive impact on some of the work we're doing. One reason is that we run a lot of tests, in which we systematically try different parameter values for aspects of our AI code (searching regions of parameter space with a GA or other optimization tool). If each test would run 10x faster, that would be a very good thing. As to how much this would accelerate our progress toward AGI, that's hard to say. Speed of running tests is only one rate-limiting factor. No amount of computer power will diminish the time it takes for humans to write and debug code, and solve the various puzzles arising in translating our mathematical/conceptual design, piece by piece, into software components My best guess is: 1) If we had vastly better CPU's and vastly more RAM, the amount of time to get to a complete working implementation of a Novamente system might be reduced to 2/3 what it is right now. 2) HOWEVER, once we get to the stage of of having a complete working implementation, the next phase is mostly a testing, tuning and teaching phase. That phase, I reckon, will be much more easily accelerable via increased CPU power. Because, it will largely be a matter of running tests to tune parameters, and the speed of doing this is roughly proportional to CPU speed (until one reaches the point where human attention to interpret the test results is the rate-limiting factor). -- Ben Goertzel --- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/?[EMAIL PROTECTED]
Re: [agi] Intelligence by definition
Yes, if you have a huge amount of space and time resources available, you can start your system with a blank slate -- nothing but a very simple learning algorithm, and let it learn how to learn, learn how to structure its memory, etc. etc. etc. This is pretty much what OOPS does, and what is suggested in Marcus Hutter's related work. It is not a practical approach, in my view. My belief is that, given realistic resource constraints, you can't take such a general approach and have to start off the system with specific learning methods, and even further than that, with a collection of functionally-specialized combinations of learning algorithms. I could be wrong of course but I have seen no evidence to the contrary, so far... *** A fixed collection of methods won't scale, - power of a method should correspond to generality (predictive power) of a pattern. The whole point of such pattern-specific & level-specific scaling of methods IS computational efficiency, - it's a lot less expensive to incrementally scale methods for individual patterns than to indiscriminately apply a fixed set of them on patterns most of which are either too complex or too simple for any given method. *** To select formulas you must have an implicit criterion, why not try to make it explicit? I don't believe we need complex math for AI, complex methods can Sorry, that was a typo, it should be "can't" be universal, - generalization is a reduction. What we need is a an autonomously scalable method. *** Well, if you know some simple math that is adequate for deriving a practical AI design, please speak up. Point me to the URL where you've posted the paper containing this math! I'll be very curious to read it ;-) *** We both know that there is no practical general AI yet, I'm trying to suggest a theoretically consistent one. Given that the whole endeavor is context-free it should ultimately be the same thing. I don't have any papers, when the theory is finished I'll write a program, not a paper. My method is ultimately simple: sequential expansion of search for correlations of sequentially increasing arithmetic power/derivation, for inputs which had above-average compression over the shorter range of search / lower arithmetic power/derivation. What's new here (correct me if I'm wrong), is how I define compression, which determines value of a pattern, & encode these patterns to preserve restorability & enable analytical comparison (between individual variable types within patterns). Both are necessary to selectively scale the search, & I don't see it in OOPS It's in my introduction, someplace, but I realize it must be mental torture to try to figure it. Why would you work on it? Only if you agree with my theoretical assumptions, I suppose, the method is uniquely consistent with them. Boris.