Dr. Jim videos - a conversation about machine intelligence
mallred
Posts: 122
Dr. Jim discusses machine intelligence in a series of 3 videos on the Machine Intelligence Technologies blog.· More in-depth than in the past.
http://machineinteltech.com/blog/blog1.php
http://machineinteltech.com/blog/blog1.php
Comments
Interesting video. Thanks for posting it.
Hope your wife is doing better.
Jim
Glad you liked the videos.
Mark
Kidney stone can cause pain like that.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
www.mikronauts.com - my site 6.250MHz custom Crystals for running Propellers at 100MHz
Las - Large model assembler for the Propeller Largos - a feature full nano operating system for the Propeller
Morpheus & Mem+ Advanced dual Propeller SBC with XMM and 256 Color VGA
Please use mikronauts _at_ gmail _dot_ com to contact me off-forum, my PM is almost totally full
Jim
I'm glad to hear your wife is doing better! If it was, indeed, a kidney stone, I can empathize fully. But, no matter what the cause, it's good to hear she's attained relief. Fortunately, our capacity to remember pain is very limited; else, I suspect, every child would be an only child!
Take care,
-Phil
Thank you all for your concern.
Mark
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Never use force, just go for a bigger hammer!
The DIY Digital-Readout for mills, lathes etc.:
YADRO
And I understood, that the Propeller doesn't have enough RAM, so it needs a memory-extension. And a permanent storage to keep that information after power down.
What a waste of time!
Nick
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Never use force, just go for a bigger hammer!
The DIY Digital-Readout for mills, lathes etc.:
YADRO
That is about what I thought when I watched it also.. bear with me here..
I'm primarily a technical guy. I believe I have a pretty good understanding of the basics, through to about the 10,000 ft view. I've been lucky enough to work with (and learn from) some exceptionally clever people. People who give a presentation that drones on and on about trying to explain the basics (badly), yet glosses over the super interesting stuff sitting about 20,000 ft up because to them, the stuff we find interesting is their bread and butter.
It used to irritate me when I sat through these presentations as I felt they were trying to teach me how to suck eggs (badly). Eventually it twigged that in most cases the people presenting were coming from a viewpoint so far removed from mine, that their grasp of the bits they were describing was extremely elementary, sometimes misguided and sometimes just plain wrong. But these people did not have to understand this stuff any more than was required to achieve the basic functionality (if at all, some have others just do it for them) they needed to do some really, really interesting things. These are the people who conceptualise the stuff we'll be using in 20 years.
You can't be good at everything, and knowing what you don't know is the most valuable thing you can learn. The stuff Dr Jim is talking about right now seems pretty basic when you look at the technology (because it is). We know there is nothing new in the base platform he is presenting, but I suspect his head is so far up in the clouds he only needs to make it do what he needs it to do to achieve a set of specific end goals.
I'm hanging on for the end game. The things he's hinted at during some of the video blogs have been so completely out there that I can't get past my 10,000 ft view to truly understand where he's coming from. I'm hoping to learn something along the way, but I'm hanging out to see what he's really going to show us.
I hope that post did not come across as arrogant.. It was intended as the opposite.. Django Reinhardt said you only stop learning when you are 6 ft under. I'm hanging on as I reckon I might learn something new, and new is good [noparse]:)[/noparse] (Shiny is better!)
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
<This space for rent>
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
JMH
Having just sat through these vidoes - twice - I feel I have to publicly agree with Nick. To NOT say something about this type of nonense would be doing a disservice to the rest of the Propeller community. I sat through these videos intending to take a few notes so that I could bring up some points to promote further discussion. This is a subject I know a little bit about (as well as simply being interested in) having done an honours thesis on neural networking. I sat through all the videos twice because I figured I MUST have missed (or perhaps misunderstood) something the first time round. Now I'm just sorry I wasted so much time.
Over 17 minutes watching a static Propeller screen (on which nothing actually happens) while listening to a voiceover by Dr Jim describing some very simplistic (and sometimes just plain wrong) notions of how the brain works - and how a machine following a program can never be intelligent (Dr Jim has clearly never heard either of chaos theory or emergent behaviour). Just why Dr Jim believes programming a Propeller is going to achieve machine intelligence when he's just finished defining machine intelligence as something that can't be achieved this way is completely inexplicable.
I also find Dr Jim's repeated insinuations (he does it again in these videos) that there is no support software, standalone debuggers or other helpful tools suitable for developing software for the Propeller annoying - but also a little sad. Even if it were true when he started out (frankly, I doubt it was the case even then) he now has access to a wealth of tools far more sophisticated than those he himself appears to be either using, selling or developing.
Ross.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Catalina - a FREE C compiler for the Propeller - see Catalina
See these old threads
Dr. Jim in Sept/Oct issue of ROBOT Magazine http://forums.parallax.com/showthread.php?p=825564
KISS OS - A new operating system for the Parallax Propeller· http://forums.parallax.com/showthread.php?p=822915
Dr. Jim in Robot Magazine··http://forums.parallax.com/showthread.php?p=809911
I note none contain any answers or information of substance, with the exception of KISS OS which states
to which I replied
That was on 22 July. No reply and no more posts to that thread.... and a new thread starts
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Links to other interesting threads:
· Home of the MultiBladeProps: TriBladeProp, RamBlade, TwinBlade,·SixBlade, website
· Single Board Computer:·3 Propeller ICs·and a·TriBladeProp board (ZiCog Z80 Emulator)
· Prop Tools under Development or Completed (Index)
· Emulators: Micros eg Altair, and Terminals eg VT100 (Index) ZiCog (Z80), MoCog (6809)
· Search the Propeller forums (via Google)
My cruising website is: ·www.bluemagic.biz·· MultiBladeProp is: www.bluemagic.biz/cluso.htm
I was trying to explain that perhaps Dr Jim was flying a lot higher than perhaps we were looking at and maybe we should wait patiently and simply see what transpires.. let's stop the mud please..
Never know.. might learn something.
I'll stop now, honest.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
<This space for rent>
"Over 17 minutes watching a static Propeller screen (on which nothing actually happens) while listening to a voiceover by Dr Jim describing some very simplistic (and sometimes just plain wrong) notions of how the brain works ..." I am interested in the "... (and sometimes just plain wrong) ..." statement. Would you be a little more specific, and identify the "... plain wrong stuff ..." And while you are at it, maybe you should describe your theory of brain functionality.
Ray
If Dr. Jim is all talk and no substance he will disappear in the wind as his words will fall on deaf ears deafened by the white noise! However if there is real substance to come let us wait and listen carefully so as to be able to hear it over our own voices!
TrapperBob
Mark
Why EVERY TIME Dr. Jim posts does everybody go against him?!?
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Toys are microcontroled.
Robots are microcontroled.
I am microcontroled.
If it's not Parallax then don't even bother. :-)
·
Mini-Din/PS2 connectors are for sale! 5 for $1! PM me if you wish to make an order.
Cheap·shipping unless specified!··········150 left!!··
So, you probably are wondering, "Mark, how the heck do you know that this even works if you have not seen it yourself?"
Ok, fair question. In an effort of full disclosure, I have seen Dr. Jim's imaging software, which also uses his machine intelligence algorithms, learn to read on a PC. I have seen it take a dark image of the ocean and identify underwater structures because of the wave patterns passing over it. I have seen many of his patents and have seen emails from the military in contact with him.
I'm not trying to prove anything here, just trying to give a little more substantiation to my claims, that's all.
In the end, I hope you join us. When this thing busts loose, we are going to need intelligent people to help with the research. I hope you will be there with us.
Thanks,
Mark
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Toys are microcontroled.
Robots are microcontroled.
I am microcontroled.
If it's not Parallax then don't even bother. :-)
·
Mini-Din/PS2 connectors are for sale! 5 for $1! PM me if you wish to make an order.
Cheap·shipping unless specified!··········150 left!!··
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Toys are microcontroled.
Robots are microcontroled.
I am microcontroled.
If it's not Parallax then don't even bother. :-)
·
Mini-Din/PS2 connectors are for sale! 5 for $1! PM me if you wish to make an order.
Cheap·shipping unless specified!··········150 left!!··
Ok, as BradC says - I'll try to be open minded about Dr Jim. He may indeed be a genius for all I know. But to answer your specific question, I'll have to point out where I think Dr Jim is wrong (and I invite Dr Jim to respond if he feels I am misrepresenting his views).
To start with, it's a fundamental misunderstanding of neurons to think that they are like digital switches and only have two states - to quote Dr Jim (from memory here) - "neurons either fire or don't fire - one or zero". This is simply not how neurons propagate information - it's the rate at which the neuron fires which does that - for example, in the visual cortex a bright light hitting the retina results in a higher rate of firing of the neurons in the visual cortex than a dim light. It's not a simple binary thing - neurons are in many ways more like analog circuits than digital ones.
Also, neurons are not passive participants in brain functioning as Dr Jim states - i.e. they are not mere "signalling stations" (I can't recall the exact the term he uses) - neurons have a complex internal "state", based on both immediate past inputs (which briefly change the ability of the neuron to "fire" - in either an "excitory" or an "inhibitory" manner) and long term past inputs (which is reflected in the strength of the synaptic connections between the neurons). The behaviour of neurons is very complex - and this complexity is essential to cognitive functioning - in fact it IS the congitive functioning. Yes, synapses are important, and the strength of the synaptic connections are probably the seat of long-term memory. But this has been known for decades - It is not (as Dr Jim implies) some new idea that people are only now starting to consider. Also, there are now known to be several other memory mechanisms at work in the brain, and at least one of these is probably based in the neurons, not in the synapes - because the synapses are known to change far too slowly to explain even short-term memory, let alone be any use for any cognitive processing which has to occur in real-time.
Of course, it is unlikely that Dr Jim is actually simulating neurons - while it's hard to tell exactly what it is he is doing, it only requires simple arithemetic to see that he cannot be doing that. Not only is the 32kb of the Prop not enough memory to perform any significant cognitive processing (Dr Jim is right about that) - neither is 16Mb. 16Mb might be enough to simulate (at a coarse level) a couple of thousand neurons - because even just a thousand neurons means upwards of a MILLION synaptic connections between them. All the synaptic connections have to be represented somehow - and this takes of the order of a megabyte no matter how cleverly you try and represent it or encode it (basic information theory says this, not me) - don't forget that each of these connections is not a simple digital value (requiring only a bit or two) - they are more like analog values (requiring potentially many bits each to represent with sufficient accuracy). But even if you can do all this properly, a couple of thousand neurons gets you about the cognitive processing power of a housefly - and how many houseflies do you know that can carry on an intelligent conversation?
So all we can really tell from the vidoes is that if neural modelling is what Dr Jim is proposing then simple information theory says he doesn't have enough memory. If he's not proposing that but instead has some other idea in mind, then he hasn't explained what it might be - or why the Propeller is a particularly good platform on which to do it.
As for my theory of brain functionality (and remember, I'm not the one who claims to have one!) - I'll happily explain mine once Dr Jim explains his. His videos certainly don't appear to explain anything.
Perhaps BradC is right and Dr Jim is soaring at 20,000 ft while the rest of us are dirtbound. I'd be willing to concede this - once Dr Jim provides some evidence.
Ross.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Catalina - a FREE C compiler for the Propeller - see Catalina
That's OK.
But what you are doing is present a still picture into a tool-box and someone talking in the background about a break-through in engine-design that makes more than 100 mpg and in which way to turn a screw. "And you only have to buy my set of screw-drivers to share this break-trough at the end of the year".
Sorry, it's *you* presenting these videos that make them look like a scam.
You have already learned that here's a rats nest of critical observers. And they want answers and explanations about principles that are new to them. Don't explain them in a 16 minutes video which way to grab a soldering iron.
Nick
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Never use force, just go for a bigger hammer!
The DIY Digital-Readout for mills, lathes etc.:
YADRO
Good Luck, Dr. Jim.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Toys are microcontroled.
Robots are microcontroled.
I am microcontroled.
If it's not Parallax then don't even bother. :-)
·
Mini-Din/PS2 connectors are for sale! 5 for $1! PM me if you wish to make an order.
Cheap·shipping unless specified!··········150 left!!··
Great! You are thinking about it. I see the gears turning.
@Nick, are you sure you watched the right set of videos?
Mark
Dr. Jim admits that his is not the only solution, but it is a start. He may not have all of it right, but that's ok. This is a good first start. I have seen things that he can do that I have not seen anyone else be able to do. So, a perfect replica of the brain is not the goal here, but trying our best to approximate our understanding of it, rather, is what we are trying to do. It can always be improved upon. Of that I am certain.
Mark
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Suzuki SV1000S motorcycle
John Abshier
Good question!
I looked at the three videos in your link. But I found no discussion about "machine intelligence". But at least the title said so.
If your question was targeted to my saying "screwdriver" and "engine": Just shift a problem or description to a different domain but keep the meaning of the words. This is a trick to sometimes understand things easier. So "screwdriver" translates to "debugger" and "zero & one" to "wich way round to turn a screw" etc.
Nick
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Never use force, just go for a bigger hammer!
The DIY Digital-Readout for mills, lathes etc.:
YADRO