On the west coast there's Shenoy's group at Stanford who are doing similar things. Much of their work is currently with monkeys and not sure if they have anything clinically available yet, but at least you can get an idea of what's currently possible:
I'm not so sure. You already need to be a domain expert to evaluate the equivalence of two ideas. Equivalence is needed to prove prior art. Why not just go the extra step and evaluate the documented effort expended to get from the previous state of things to the new claimed state, along with all reasonable missteps along the way. An expert could easily spot bullshit claimed steps or useless missteps.
All I'm basically pointing out here is that the current system doesn't provide rewards in proportion to work expended. Instead it prioritizes being there first. Land grabs are useful when it's difficult for effective sharing of a common resource. An idea can be copied and shared with little penalty. Why would you want to limit that advantage? Except of course to make sure idea generation is properly incentivized.
Not quite sure why the focus is so much on prior art. Sure an idea that has prior art can serve as proof of the obviousness of something, but at some point in time someone is, in fact, the first to propose some blend of existing ideas and call it new. Do we really need to permit that person a patent? For most low lying fruit this is basically the equivalent of a land grab.
The way I see it patents need to be granted in proportion to the amount of work required to explore the possible parameter space to find that new unique useful combination. It shouldn't just be about being first to something, it should be about expending a lot of effort to get there. And the reward for that should be temporary and in proportion to that amount of work. Shouldn't it be fairly easy to offer up objective proof of spending that effort (and have it peer reviewed)?
The process of figuring this out isn't going to occur magically. You need to test out your models at the systems level, with all the components working together. The more powerful the hardware we have to do this the more we can test and refine our models of how the brain achieves the same thing. This is both true if you're trying to model existing neuro architectures (like BU is) or if you're modeling evolutionary approaches like you describe above.
These memristive neuromorphic architectures hold the promise to get us orders of magnitude more processing speed while also keeping power levels low.
It can and is being designed for that use but I believe there have been problems with reliability of individual memristor units.. However in a neuromorphic design (non-Von Neumann architecture) you only need a certain percentage of the units to be reliable as the information is highly distributed and fault tolerant. Think of the massive cell death that occurs in Alzheimer's disease, yet patients are still fairly normal well into that process.
The other main advantage is that you can represent a single synapse with a single memristor which can be smaller than a single transistor.
I mean arbitray in the sense that if you were designing a hippocampus to support sequence retention you could design it either way. The connections between cells that represent elements in the sequence need to be re-enforced by repetitive activation. I guess this would be what the rat is doing while resting.
Then again, there's spike timing dependent plasticity (STDP) to consider. I'm not sure how this is effected by the sequence being played in reverse. The hippocampus does seem to be multimodal (learning, recall, others?) depending on the frequency it is being globally modulated by (as well as by neurotransmitter modulation).
Perhaps I'm wrong about hippocampal neurons needing re-enforcement at this point. The playback in reverse may soley serve a purpose in hippocampal-cortical interactions. Perhaps in forming the gestalt representations I mentioned. I believe these exist in the neighboring entorhinal region (combinations of proximally occuring object representations). The fact that the neurons are firing rapidly in this mode leads me to think the purpose is the training of a grouped representation.
This kind of temporal difference learning is the job of the cerebellum (short time scales) and the basil ganglia (arbitrary time scales). I would bet that the fact that the sequence is replayed backwards in this case is just arbitrary. The idea is to re-enforce the connections between the neurons that represent the learned sequence, as well possibly to train cortical neurons in gestalt like representions of the sequence.
You are right in that this type of sequencial inforamtion is capable of encoding causation. These hippocampal sequence representations are also relayed to the cortext, which is where plans are represented and executed.
The cortext is where this sequencial information is filtered. It's much slower to learn and tends to filter out only the most consistent sequencial patterns and their constituents. In this way it compresses the sensory input from the environment (as well as motor command feedback) into the most statistically significant causal relations. These relations along with goal state representations form the building blocks for plans.
When I say neuroscientist I probably should have been more specific: a computational neuroscientist. I spend most of my time investigating the 'mysteries of the mind' as you call them. Since I'm building computer models I'm actually testing my hypotheses (unlike ID advocates), and so far I've not found any need to postulate the existence of a 'spirit' to fill in gaps in understanding.
The type of experience you describe can be easily accounted for by your cortex entering into a constellation of activation not usually possible with normal interaction with your environment. This can be achieved with the use of drugs, epilepsy, strokes, oxygen deprivation, meditation, etc.
Being a neuroscientist myself, I'm a always a little ticked off that these ID advocates attribute such mysterious qualities to the concept of intelligence. At it's core, intelligence is the action of copying and blending directed by reinforcement learning and random circumstance.
If you compare this to evolution you have copying through reproduction, blending through sex and symbiosis, natural selection as a form of reinforcement learning, and of course mutation to equate to circumstance. It's probably not a perfect analogy but it's still close. In fact, one way of lookinng at the brain is that it affords us the ability to carry out the proccess of evolution during our own lifetimes, rather than having to wait over generations.
So anyway, to say that life is too complex for evolution to have produced it is in my book very much the same thing as saying life is too complex for intelligence to have produced it. The latter is falsifiable since intelligence has already created artificial forms of life, not to mention is currently toying around with creating new forms of biological life.
If you don't buy my definition of intelligence, I'd be happy to debate it here.
All very true, but if the question were phrased: "What's the simplest generating rule for this sequence?" There would be a lot fewer answers and quite likely one unique answer. The emphisis should be on compressing the available information and not on predicting future information.
You fogot the.1% of posts predicting the percentage of each category of post. Not to mention the % in the category of posts pointing out that you forgot the former category.
In response to edge.org's question "What do you believe is true even though you cannot prove it?" Jordon Pollock gave this response which I think is particulary relevant:
JORDAN POLLACK
Computer Scientist, Brandeis University
I believe that that systems of self-interested agents can make progress on their own without centralized supervision.
There is an isomorphism between evolution, economics, and education. In economics, the supervisor is a central government or super rich investor, in evolution, it is the "intelligent designer", and in education, its the teacher or outside examiners. In economic systems, despite an almost religious belief in Laissez-Faire and incentive-based behavior, economic systems are prone to winner-take-all phenomena and boom-bust cycles. They seem to require benevolent regulation, or "managed competition" to prevent the "rich get richer" dynamic leading to monopoly, which leads inevitably to corruption and kleptocracy. In evolution, scientists reject the intelligent designer as a creationist ruse, but so far our working models for open-ended evolution haven't worked, and prematurely convergence to mediocrity. In education, evidence of auto-didactic learning in video-games and sports is suppressed in academics by top-down curriculum frameworks and centralized high-stakes testing.
If we did have a working mechanism design which could achieve continuous progress by decentralized self-interested agents, it would settle the creationist objection as well as apply to the other fields, leading to a new renaissance.
Just because something is beyond our current level of comprehension does not necessitate that something more intelligent and powerful than us had to have created it. In fact, it's rather easy to create enormously complex systems from a set of simple rules applied iteratively. Complexity is cheap. Check out Stephen Wolfram's new book if you're interested. You also might check out books on chaos theory or self-organizing criticality.
Oh and incidentally, having a willingness to change and having absolute faith in something are to a large extent mutually exclusive.
You might want to check to see if it actually uninstalled completely. Fire up your task manager and make sure you know what all those programs are there for. Also check in your Common Files dir and do the same.
When I uninstalled it I had to physically clean out my registry, delete the program (in the Common Files subdir of Program Files) as well as kill the task. As far as I can tell the uninstaller did nothing.
You have an interesting but highly tedious approach to AI there. But even if you were to complete such an ambitious project your resulting database would not be inherently inteligent. You are still relying on humans to filter the world and extract the relevant knowledge encoded into the database.
This is no different than current attempts at AI's to play Chess, Go, etc. Human experience (with a lot of lookahead searching) is hard coded into the system. Becuause that information is hard coded a human will always be able to adapt to and beat such a system.
A truely inteligent system is capable of extracting its own knowledge base from its environment and using that information to its benefit (that benefit being defined by the designer of the AI). This will enable it to adapt to changing conditions and alleviate the need for any furthur human input.
Wow...Hefty Smurf..long time no see... I've since graduated (almost didn't pass as a result of quake) and have a real life as well, but I still play q1 occationally. Believe it or not there's still a fairly competetive community out there...thanks in part to me I guess for continuing to run both the east and west teamplay.net servers.
DM2 is not a bad 2on2 level..tho i'd rather play dm3 2on2. DM6 is the perfect size...with rl jumping you can really move around that level fast. The reason ID made DM1 was they felt they needed a level without an rl or shaft so ppl would learn to use other weapons...little did they know how sucky those other weapons were in comparison and hence everyone's hatred of it.
Re:5 years since the release of Quake means...
on
Five Years of Quake
·
· Score: 2
With the release of the source code a while back I've long since had quake1 running stably under linux not to mention much faster than under windows.. Nvidia's new linux drivers are hands down faster than they're windows drivers.
Both DM2 and DM4 have major flaws for competive play. In DM2 it's way too easy to run the rl's and DM4 has a dead end that severly slows down a game. DM6 is quite possibly the perfect 1on1 level. I've yet to see a 1on1 level even come close in terms of balance and speed. DM3 is best team game level I've come across. DM6 and 3 are what is keeping q1 alive today.
What made Quake really different than say doom or wolfenstien was the multiplayer code. After playing quake online with other ppl I've have since been unable to play a single player game. The amount of fun to be had trying to outsmart some static computer algorithm pales in comparison to trying to adapt to other real players. Quake was the first game to have infinite replayability. As proof of that I still play the original quake1 today avidly.
I'm not too keen on anything else Id or anyone else has produced in the fps area since making the game look nice has since been the primary goal of those game developers. And making it look nice and having it playable are contradictory imho. Take for example the way good q3 players play: gl_picmip 5...the game looks like ass..whereas with just plain old glquake where you have a fixed set of colors from an 8 bit palet you have a much cleaner looking game and hence it's easier to interact with.
Slashdot Mirror
On the west coast there's Shenoy's group at Stanford who are doing similar things. Much of their work is currently with monkeys and not sure if they have anything clinically available yet, but at least you can get an idea of what's currently possible:
http://www.stanford.edu/~sheno...
You can build a fairly large simulation that runs on modern inexpensive graphics hardware. Learning how to program those is where I'd start.
I'm not so sure. You already need to be a domain expert to evaluate the equivalence of two ideas. Equivalence is needed to prove prior art. Why not just go the extra step and evaluate the documented effort expended to get from the previous state of things to the new claimed state, along with all reasonable missteps along the way. An expert could easily spot bullshit claimed steps or useless missteps.
All I'm basically pointing out here is that the current system doesn't provide rewards in proportion to work expended. Instead it prioritizes being there first. Land grabs are useful when it's difficult for effective sharing of a common resource. An idea can be copied and shared with little penalty. Why would you want to limit that advantage? Except of course to make sure idea generation is properly incentivized.
Not quite sure why the focus is so much on prior art. Sure an idea that has prior art can serve as proof of the obviousness of something, but at some point in time someone is, in fact, the first to propose some blend of existing ideas and call it new. Do we really need to permit that person a patent? For most low lying fruit this is basically the equivalent of a land grab.
The way I see it patents need to be granted in proportion to the amount of work required to explore the possible parameter space to find that new unique useful combination. It shouldn't just be about being first to something, it should be about expending a lot of effort to get there. And the reward for that should be temporary and in proportion to that amount of work. Shouldn't it be fairly easy to offer up objective proof of spending that effort (and have it peer reviewed)?
The process of figuring this out isn't going to occur magically. You need to test out your models at the systems level, with all the components working together. The more powerful the hardware we have to do this the more we can test and refine our models of how the brain achieves the same thing. This is both true if you're trying to model existing neuro architectures (like BU is) or if you're modeling evolutionary approaches like you describe above.
These memristive neuromorphic architectures hold the promise to get us orders of magnitude more processing speed while also keeping power levels low.
It can and is being designed for that use but I believe there have been problems with reliability of individual memristor units.. However in a neuromorphic design (non-Von Neumann architecture) you only need a certain percentage of the units to be reliable as the information is highly distributed and fault tolerant. Think of the massive cell death that occurs in Alzheimer's disease, yet patients are still fairly normal well into that process.
The other main advantage is that you can represent a single synapse with a single memristor which can be smaller than a single transistor.
I mean arbitray in the sense that if you were designing a hippocampus to support sequence retention you could design it either way. The connections between cells that represent elements in the sequence need to be re-enforced by repetitive activation. I guess this would be what the rat is doing while resting.
Then again, there's spike timing dependent plasticity (STDP) to consider. I'm not sure how this is effected by the sequence being played in reverse. The hippocampus does seem to be multimodal (learning, recall, others?) depending on the frequency it is being globally modulated by (as well as by neurotransmitter modulation).
Perhaps I'm wrong about hippocampal neurons needing re-enforcement at this point. The playback in reverse may soley serve a purpose in hippocampal-cortical interactions. Perhaps in forming the gestalt representations I mentioned. I believe these exist in the neighboring entorhinal region (combinations of proximally occuring object representations). The fact that the neurons are firing rapidly in this mode leads me to think the purpose is the training of a grouped representation.
This kind of temporal difference learning is the job of the cerebellum (short time scales) and the basil ganglia (arbitrary time scales). I would bet that the fact that the sequence is replayed backwards in this case is just arbitrary. The idea is to re-enforce the connections between the neurons that represent the learned sequence, as well possibly to train cortical neurons in gestalt like representions of the sequence.
You are right in that this type of sequencial inforamtion is capable of encoding causation. These hippocampal sequence representations are also relayed to the cortext, which is where plans are represented and executed.
The cortext is where this sequencial information is filtered. It's much slower to learn and tends to filter out only the most consistent sequencial patterns and their constituents. In this way it compresses the sensory input from the environment (as well as motor command feedback) into the most statistically significant causal relations. These relations along with goal state representations form the building blocks for plans.
When I say neuroscientist I probably should have been more specific: a computational neuroscientist. I spend most of my time investigating the 'mysteries of the mind' as you call them. Since I'm building computer models I'm actually testing my hypotheses (unlike ID advocates), and so far I've not found any need to postulate the existence of a 'spirit' to fill in gaps in understanding.
The type of experience you describe can be easily accounted for by your cortex entering into a constellation of activation not usually possible with normal interaction with your environment. This can be achieved with the use of drugs, epilepsy, strokes, oxygen deprivation, meditation, etc.
Being a neuroscientist myself, I'm a always a little ticked off that these ID advocates attribute such mysterious qualities to the concept of intelligence. At it's core, intelligence is the action of copying and blending directed by reinforcement learning and random circumstance.
If you compare this to evolution you have copying through reproduction, blending through sex and symbiosis, natural selection as a form of reinforcement learning, and of course mutation to equate to circumstance. It's probably not a perfect analogy but it's still close. In fact, one way of lookinng at the brain is that it affords us the ability to carry out the proccess of evolution during our own lifetimes, rather than having to wait over generations.
So anyway, to say that life is too complex for evolution to have produced it is in my book very much the same thing as saying life is too complex for intelligence to have produced it. The latter is falsifiable since intelligence has already created artificial forms of life, not to mention is currently toying around with creating new forms of biological life.
If you don't buy my definition of intelligence, I'd be happy to debate it here.
All very true, but if the question were phrased: "What's the simplest generating rule for this sequence?" There would be a lot fewer answers and quite likely one unique answer. The emphisis should be on compressing the available information and not on predicting future information.
You fogot the .1% of posts predicting the percentage of each category of post. Not to mention the % in the category of posts pointing out that you forgot the former category.
Just because something is beyond our current level of comprehension does not necessitate that something more intelligent and powerful than us had to have created it. In fact, it's rather easy to create enormously complex systems from a set of simple rules applied iteratively. Complexity is cheap. Check out Stephen Wolfram's new book if you're interested. You also might check out books on chaos theory or self-organizing criticality.
Oh and incidentally, having a willingness to change and having absolute faith in something are to a large extent mutually exclusive.
In its place they found a mixture of green dish detergent and some glitter.
Even Quake1 had it: sv_maxspeed
It also had to be set on the server side unless you were running a mod that allowed the change from the client side.
how far do you think we would of evolved if say everyone had food and didn't have to worry about predators?
Sounds about like living in the United States to me.
You might want to check to see if it actually uninstalled completely. Fire up your task manager and make sure you know what all those programs are there for. Also check in your Common Files dir and do the same.
When I uninstalled it I had to physically clean out my registry, delete the program (in the Common Files subdir of Program Files) as well as kill the task. As far as I can tell the uninstaller did nothing.
You have an interesting but highly tedious approach to AI there. But even if you were to complete such an ambitious project your resulting database would not be inherently inteligent. You are still relying on humans to filter the world and extract the relevant knowledge encoded into the database.
This is no different than current attempts at AI's to play Chess, Go, etc. Human experience (with a lot of lookahead searching) is hard coded into the system. Becuause that information is hard coded a human will always be able to adapt to and beat such a system.
A truely inteligent system is capable of extracting its own knowledge base from its environment and using that information to its benefit (that benefit being defined by the designer of the AI). This will enable it to adapt to changing conditions and alleviate the need for any furthur human input.
Wow...Hefty Smurf..long time no see... I've since graduated (almost didn't pass as a result of quake) and have a real life as well, but I still play q1 occationally. Believe it or not there's still a fairly competetive community out there...thanks in part to me I guess for continuing to run both the east and west teamplay.net servers.
DM2 is not a bad 2on2 level..tho i'd rather play dm3 2on2. DM6 is the perfect size...with rl jumping you can really move around that level fast. The reason ID made DM1 was they felt they needed a level without an rl or shaft so ppl would learn to use other weapons...little did they know how sucky those other weapons were in comparison and hence everyone's hatred of it.
With the release of the source code a while back I've long since had quake1 running stably under linux not to mention much faster than under windows.. Nvidia's new linux drivers are hands down faster than they're windows drivers.
Both DM2 and DM4 have major flaws for competive play. In DM2 it's way too easy to run the rl's and DM4 has a dead end that severly slows down a game. DM6 is quite possibly the perfect 1on1 level. I've yet to see a 1on1 level even come close in terms of balance and speed. DM3 is best team game level I've come across. DM6 and 3 are what is keeping q1 alive today.
What made Quake really different than say doom or wolfenstien was the multiplayer code. After playing quake online with other ppl I've have since been unable to play a single player game. The amount of fun to be had trying to outsmart some static computer algorithm pales in comparison to trying to adapt to other real players. Quake was the first game to have infinite replayability. As proof of that I still play the original quake1 today avidly. ...the game looks like ass..whereas with just plain old glquake where you have a fixed set of colors from an 8 bit palet you have a much cleaner looking game and hence it's easier to interact with.
I'm not too keen on anything else Id or anyone else has produced in the fps area since making the game look nice has since been the primary goal of those game developers. And making it look nice and having it playable are contradictory imho. Take for example the way good q3 players play: gl_picmip 5