Airplanes don't have feathers

2011-06-10T17:28:00.000-07:00

Modern Airlanes don't have feathers. But they still can fly.
When copying nature, you don't need to copy it exactly as it is.
The exact simulation of (human) nerve cells is dispensable.
It's enough to understand and build the principles behind.
The first attempts to fly with feathers did fail.

What fires together, wires together

2010-06-03T14:46:00.000-07:00

This is the key to all AI, so please never forget Hebb's postulate:
"What fires together wires together".

And also remember:
cognition is always re-cognition

Move across the horizon

2009-09-03T17:42:00.000-07:00

What would you do with a peta-flop?

How would you move windows on the screen if you had peta-flops?

Transparency? Themes? HD? 3D? Any other ideas?

Yes, of course ...

Today, we have tera flops. 10¹² Floating Point Operations Per Second. In our graphics card. On the playstation's cell processor.

In the near future we will have peta flops. What would you wanna do with it? Move windows on the screen for a better user experience ?

Imagine, someone offered a ship to Christopher Columbus, that could sail across the horizon. What would Columbus like to do with it ?

Hiss more beautiful colored sails? Stay in harbour and try to make the anchors more compatible with the ground?

What did we expect computers to be?

THINKING machines.

Think about it.

Detlef

Some thoughts on neuro-sciences. From IT to KM

2008-03-29T19:03:00.000-07:00

Data is not information, information is not knowledge, knowledge is not wisdom, wisdom is not truth.

From my first conact with computers (in 1986), I was interested in artificial intelligence (AI). I asked myself, if this was possible. I started with programming and I am now very busy with my products and customers.

The key to AI is, how the brain stores it's information.

We all know some sudden memories from early childhood, that come into our mind. These days I listened to music I haven't heard for 40 years. I didn't remember that I know it, but listening to it again, I can remember every single note.

Both examples show, that the brain must be able to store a huge amount of information. It seems to store everything ever seen, listened and felt in a human life. But how much is this in GigaBytes, TeraBytes or even PetaBytes ?

We know that even PetaBytes might not be enough.
Is the storage system some kind of holistic? Or associative? Or well compressed?

A hologram stores a picture. When cut into two parts, every part will show the entire picture. The brain seems to store it's information somehow holistic. Everyday thousands of neurons die without loosing too much information.

Associative storage algorithms can eliminate redundant data by storing references on the redundant entries. The brain seems to store it's information in an associative manner.

We know many good compression algorithms. I don't think that the secret of the brain lays in a magic kind of data compression

>>>>>> Neurosciences <<<<<<

In the age of Christoph Columbus, everyone believed in the flat earth myth. Columbus sailed westwards and discovered america. He could have sailed in any direction and he would have found something!

Today we have models from the smallest particles to the biggest thing (the universe). We have an imagination from the beginning of time (big bang) and we have sequenced the genome.

But the human brain is a total terra incognita.

A man who travels in this area will find something, same as Columbus did.

Is the mind a function of the brain?
Buddhists like Patanjali and some scientists from the 1970's asked this question. If this can be answered with yes, AI is possible. If not, it's time to get religious...

I think, the key to find out more, is to understand, how the brain stores information. A later step will be to see how the information is processed. Pattern-matching and pattern-recognition are for me the first things to explore...

We can see electrons through the electron microscope and we know the structure of the brain. It is complex and big but also simple, because it is built with one single type of cells, the neuron. But we have no idea how the neuron works and how it stores information.

I think, this will be the next BIG THING. It will lead to many new technologies and answer many old questions.

Concepts and theories will be proved by working computer models and programs. Neuro-Scientists and programmers will influence each other. Data-Mining and pattern-matching is a very first and practical approach in this area. (I'm into it with my FuzzyDupes software.) Neural networks didn't lead too far. But SOM's (Self Organizing Maps, a specialized application of neural network algorithms) allow automatic classification systems and are the next step to do.

We are moving from IT (Information Technology) to KM (Knowledge Management).

Tonight, I had these ideas very clearly, so I had to write them down.

Detlef

Kroll-Software

Airplanes don't have feathers

What fires together, wires together

Move across the horizon

What would you do with a peta-flop?

Some thoughts on neuro-sciences. From IT to KM