GENIUS FORUMS

prince wrote:
Get back to me when you have something I don't know, which will be never.

Wow! That's wild that you know everything.

Here is a question I have for you:

I'm using Joone to create a neural network that differentiates faces. I'm trying to use a single image input layer, a Sanger synapse to a WTA layer and finally a file output layer. The problem is, when I train the network it stalls at around 120 epochs even when the epochs are set higher. I don't know if the problem is in the particular kind of hidden layer or synapse I'm using. The problem seems to be that a standard desktop PC isn't powerful enough to handle this particular network design, or maybe there is a flaw in my understanding of the logic of Sanger and a Kholberg synapse in combination with a gaussian map would be better. What do you think?

Did you gloss over everything I have ever written on this forum?

You write stuff on this forum?

Oh, I found your "writings" Prince: search.php?keywords=&terms=all&author=p ... mit=Search

I wouldn't exactly call any of these significant, at least, not the first 5 or so pages worth.

Animus wrote:that you know everything.

Here is a question I have for you:

I'm using Joone to create a neural network that differentiates faces. I'm trying to use a single image input layer, a Sanger synapse to a WTA layer and finally a file output layer. The problem is, when I train the network it stalls at around 120 epochs even when the epochs are set higher. I don't know if the problem is in the particular kind of hidden layer or synapse I'm using. The problem seems to be that a standard desktop PC isn't powerful enough to handle this particular network design, or maybe there is a flaw in my understanding of the logic of Sanger and a Kholberg synapse in combination with a gaussian map would be better. What do you think?

So you are going through all this blood, sweat and tears to create some vulgar, poxy little image on a silly little screen,

yet you don't think that the universe itself was created?

dreams? sub to sub contact.

that was funny prince. those who claim they know everything truly know nothing. arrogance is ignorance.

i agree i know i no nothing im just giving ideas out because my mind is full of these types of things.

prince wrote:So you are going through all this blood, sweat and tears to create some vulgar, poxy little image on a silly little screen,

yet you don't think that the universe itself was created?

I'm going through this trouble so that I can learn what I don't already know. But since you know everything, can you help me with this one?

I've read through the entire Joone: A Complete Guide by Paolo Marrone but it didn't specifically help with this question.

On a more introductory level; I'm still encountering cognitive difficulties with understanding mathematical truth tables. My maths skills have always been sub-graduate level. Any help would be appreciated.

Animus wrote:I'm going through this trouble so that I can learn what I don't already know. But since you know everything, can you help me with this one?

I don't know every empirical detail, obviously, what I'm pointing out to you is that what you "don't already know" (about this specific area for example ) is what is known by a higher consciousness than human. It's already there, all of it, you are just uncovering it. It's self-evident, all technology that is possible in the chemical realm is there in some form, it's already been done by a creator.

It's a bit like a child playing in a sandpit and making little castles, totally oblivious to the fact that the sandpit, the sand, the water and it's own human form with limbs etc etc was already put there to enable the child to play in it.

prince wrote:I don't know every empirical detail, obviously, what I'm pointing out to you is that what you "don't already know" (about this specific area for example ) is what is known by a higher consciousness than human. It's already there, all of it, you are just uncovering it. It's self-evident, all technology that is possible in the chemical realm is there in some form, it's already been done by a creator.

It's a bit like a child playing in a sandpit and making little castles, totally oblivious to the fact that the sandpit, the sand, the water and it's own human form with limbs etc etc was already put there to enable the child to play in it.

And you are not that higher consciousness Prince. Allow me to relate to you the lyrics of a song by Collective Soul that addresses this issue:

"God may be good and God may be great, but God you'll never be."

Clearly you do not know everything. What you do know is a singular philosophical concept about the nature of God and reality. That has nothing to do with DMT or the clinical observations of NDErs. There is nothing about your knowledge that crosses over to what we are talking about here. Apparently, you can't channel this higher wisdom to answer any of my questions, so here's a tip. Buy a fuckin' book.

prince wrote:

Animus wrote:I'm going through this trouble so that I can learn what I don't already know. But since you know everything, can you help me with this one?

I don't know every empirical detail, obviously, what I'm pointing out to you is that what you "don't already know" (about this specific area for example ) is what is known by a higher consciousness than human. It's already there, all of it, you are just uncovering it. It's self-evident, all technology that is possible in the chemical realm is there in some form, it's already been done by a creator.

It's a bit like a child playing in a sandpit and making little castles, totally oblivious to the fact that the sandpit, the sand, the water and it's own human form with limbs etc etc was already put there to enable the child to play in it.

The only way in which this applies to DMT and perhaps the way you intended, is that in order for a chemical like DMT to act on receptors in the brain, there has to be receptors tuned to that chemical, but not that chemical specifically.

For example, sugar interacts with various taste cells and produces the phenomenal sensation of "sweet", however there are several other substances that produce similar effects by interacting with the same receptor sites, artificial sweeteners accomplish this task. However, artificial sweeteners are - in many ways - different from the natural sugars found in fruits. Some of these substances produce the phenomenal sensation of "sweet" but also decrease the potential for developing Type II Diabetes or increase the chance of developing carcinoma.

In this same manner, there are receptors in the brain responding to nicotine - a substance that has virtually no positive effects for the brain/body/spirit. In short, just because the brain contains receptors reactive to a certain chemical, does not mean they evolved for that purpose, or that consuming that chemical is a good idea. It is more likely the case that you are exploiting a receptor site that has evolved for something else.

By the way, its probably not a good idea to exploit receptors as such unless you know exactly what you are doing. Receptor activation instigates metabolic processes which alter the structure of the brain.

One of example of this would be the aeropsia I perceive after years of smoking marijuana. If I didn't know better I might think I was slipping into an alternative reality, or perceiving content within this reality that was unlocked by my philosophical mind. Nope, just too much pot.

Animus wrote:I'm using Joone to create a neural network that differentiates faces. I'm trying to use a single image input layer, a Sanger synapse to a WTA layer and finally a file output layer. The problem is, when I train the network it stalls at around 120 epochs even when the epochs are set higher.

By "it stalls" do you mean the program or the computer? (I guess you mean the program stalls)

Could be a error in the script if it stalls consistently at 120.

1456200423 wrote:

Animus wrote:I'm using Joone to create a neural network that differentiates faces. I'm trying to use a single image input layer, a Sanger synapse to a WTA layer and finally a file output layer. The problem is, when I train the network it stalls at around 120 epochs even when the epochs are set higher.

By "it stalls" do you mean the program or the computer?

Sorry, the program stalls. Task manager shows that Java VM is using 50% of the CPU resources and roughly 250MB memory, while Joone itself is using 0% of CPU and 150MB of memory. Honestly, I think its a problem with how I have the network setup. I haven't fixed all of the input images for size or done any color adjustments on them. I've read that an 80 node network consisting of sigmoid layers will do, if the input images are scaled down and certain color adjustments are applied. With only a grey scale set of input images this process apparently took hours to run. I was using images with depths of color to the billions and all were different sizes. However, I want the network to be able to handle these as well!

I think I might have to write some extra java outside of Joone to be able to handle these images. Perhaps a scalable network is all I need, something that I can feed size variables into on-the-fly.

All of this is probably well beyond my capabilities anyway. I was taking a jab at Prince for claiming to know everything.

Animus wrote:I think I might have to write some extra java outside of Joone to be able to handle these images. Perhaps a scalable network is all I need, something that I can feed size variables into on-the-fly.

All of this is probably well beyond my capabilities anyway. I was taking a jab at Prince for claiming to know everything.

Have you tried something simple, just to get the feel for the program?
Great find btw. I am checking it (Joone) out.
Always wanted a self learning chess engine... :-P

Animus wrote:I was taking a jab at Prince for claiming to know everything.

I would think that someone with your user name would believe in genetic memory. :-P
As a engineered species we have our intelligence deliberately *caped* well below our real potential.

1456200423 wrote:Have you tried something simple, just to get the feel for the program?
Great find btw. I am checking it (Joone) out.
Always wanted a self learning chess engine... :-P

I had a simple network setup using a imageinput layer to a linear layer to a sigmoid layer and finally an imageoutput layer. In that case I got the output of images that looked like gaussian color maps. I thought if I inserted a WinnerTakesAll layer I could get a text output indicating one or the other pattern had been recognized. But this stalled the network as well. I tried changing it up, adding the Sanger synapse to try to reduce the amount of input, still no luck. I think either I'm going at it completely wrong or my desktop can't handle such processes. More than likely I'm just doing it wrong, but this is all about the learning experience and not about making a functional application. Perhaps I'll find a use - like a chess program - sometime in the future, but for now I'm just trying to get my head around the detailed logic of artificial neural networks. I've done a fair amount of reading of the subject, but Joone allows me to get some practical experience as well.

I hope you enjoy the software, I'm using the GUI, though its not included in the public license and probably taking the easy way out. Please show us if you get anything out of it, I know some really cool SOM (Self-organizing Map) applications can be made with ANNs.

Animus wrote:I had a simple network setup using a imageinput layer to a linear layer to a sigmoid layer and finally an imageoutput layer. In that case I got the output of images that looked like gaussian color maps. I thought if I inserted a WinnerTakesAll layer I could get a text output indicating one or the other pattern had been recognized.

If you are after facial recognition, maybe try to make the program to test for deviation from the default template (facial golden ratio) feature construction. That's how human brain does it...
Should use simple black on white line drawing pictures at first,(to train the edge finding algorithm?) not color photos. :-)

http://www.youtube.com/watch?v=GO3o9drC1mQ

technology: anti-philosophy. one ponders our beginning and our existence, the other tries to make it easier, but will only produce the opposite. if only ted kaczynski didn't go around bombing... technology will lead to overwhelming power, and an equilibrium force to counteract this.

Animus wrote:One of example of this would be the aeropsia I perceive after years of smoking marijuana. If I didn't know better I might think I was slipping into an alternative reality, or perceiving content within this reality that was unlocked by my philosophical mind. Nope, just too much pot.

Well I think it says it all right there. The brain can only take so much damage, it's a delicately designed thing.

You are constantly using it up, it's a finite object, created for a purpose, no different from a car or a loaf of bread.

prince wrote:Well I think it says it all right there. The brain can only take so much damage, it's a delicately designed thing.

Of course it is, evolution designed it. You know, an excellent book on the haphazard construction of the human mind is Kluge by Gary Marcus, subtitled "The Haphazard Construction of the Human Mind". It very much delves into the "delicate design" of the brain by evolution and not an intelligent designer.

You are constantly using it up, it's a finite object, created for a purpose, no different from a car or a loaf of bread.

Well, aside from the aeropsia and extreme loss of memory in respect to dates and stories, I still maintain supreme problem solving skills and a memory for numbers that is unsurpassed by anyone I personally know - but not as good as the likes of Daniel Tammet.


On the weekend, my roommate and I purchased and downloaded the game "TV Show King" from the Playstation Store on my PS3. It is a typical trivia game in which contestants select one of four answers to a given question by pressing the corresponding button on the controller. In one of the playable modes called "Face Off" two players go head-to-head as the name might suggest. My roommate challenged me to a round of face-off to see who has the superior knowledge. Fine I thought, but we encountered a serious problem; she was able to answer the questions before I was half finished reading it and on some occasions she answered correctly prior to my even becoming conscious of the screen transition.

So fine, my memory is poor in respect to dates and the details of movies or events in my life and my consciousness is significantly delayed, but hey, I can still figure out why all of this happens.

It's mainly short-term memory that is effected by drugs.

And evolution didn't create the brain, a designer did. It's an organic machine, like it or lump it.

prince wrote:And evolution didn't create the brain, a designer did.

An intelligent designer?