|
|
Post by Baron von Lotsov on Nov 19, 2022 15:19:28 GMT
For a long time we have had a fairly crude idea about neural networks and so it may explain why our AI, which attempts to model the neurons in the brain, tends to be a tricky thing to get working and suffers from problems. What can easily happen in a neural net is where you sum many inputs it can saturate the range of values it is designed to work with. The standard way is to use a sigmoid function to get around this, but this is not the way the brain works. The brain has two types of neuron which when combined together enable the brain to deal with logarithmic values rather than simple values, so it can for example interpret the information coming from the eye which can detect many orders of magnitude in light intensity.
A further understanding from this is that these two types of neuron in the real world represent the difference between detail and overall picture. It got me wondering if this could explain what so-called autism is. Perhaps autistic people have more of the detail type of neuron than those that deal with the overall picture. The behaviour is consistent and it seems reasonable.
|
|
|
|
Post by Orac on Nov 20, 2022 18:06:55 GMT
I think many natural phenomena and the link to their consequences follow a similar pattern. The Earth quake metric follows the same notion and that is based on a consequential type logic and so this is perhaps intrinsic to the world and its consequences and it is no surprise that biological organisms would only feed the level of detail needed to track consequence / important change
|
|
|
|
Post by Baron von Lotsov on Nov 20, 2022 21:56:23 GMT
I think many natural phenomena and the link to their consequences follow a similar pattern. The Earth quake metric follows the same notion and that is based on a consequential type logic and so this is perhaps intrinsic to the world and its consequences and it is no surprise that biological organisms would only feed the level of detail needed to track consequence / important change There was a theory kicking about a long time ago that the human species has adapted to live in groups and as such the groups are connected to survival so it can influence the DNA. It is supposed that the DNA works better if it produces a few of the more scientific detail types and more of the other type, so like within each group you'd only need one or two to do the figuring out and inventing on how to catch food, build shelters etc and then the rest perform other functions. Anyway, it was just a hypothesis I think. |
|
|
|
Post by borchester on Nov 21, 2022 0:11:45 GMT
For a long time we have had a fairly crude idea about neural networks and so it may explain why our AI, which attempts to model the neurons in the brain, tends to be a tricky thing to get working and suffers from problems. What can easily happen in a neural net is where you sum many inputs it can saturate the range of values it is designed to work with. The standard way is to use a sigmoid function to get around this, but this is not the way the brain works. The brain has two types of neuron which when combined together enable the brain to deal with logarithmic values rather than simple values, so it can for example interpret the information coming from the eye which can detect many orders of magnitude in light intensity.
A further understanding from this is that these two types of neuron in the real world represent the difference between detail and overall picture. It got me wondering if this could explain what so-called autism is. Perhaps autistic people have more of the detail type of neuron than those that deal with the overall picture. The behaviour is consistent and it seems reasonable.
What logarithmic base do the neurons use? |
|
|
|
Post by Baron von Lotsov on Nov 21, 2022 20:12:22 GMT
For a long time we have had a fairly crude idea about neural networks and so it may explain why our AI, which attempts to model the neurons in the brain, tends to be a tricky thing to get working and suffers from problems. What can easily happen in a neural net is where you sum many inputs it can saturate the range of values it is designed to work with. The standard way is to use a sigmoid function to get around this, but this is not the way the brain works. The brain has two types of neuron which when combined together enable the brain to deal with logarithmic values rather than simple values, so it can for example interpret the information coming from the eye which can detect many orders of magnitude in light intensity.
A further understanding from this is that these two types of neuron in the real world represent the difference between detail and overall picture. It got me wondering if this could explain what so-called autism is. Perhaps autistic people have more of the detail type of neuron than those that deal with the overall picture. The behaviour is consistent and it seems reasonable.
What logarithmic base do the neurons use? He does not say and the paper is unavailable. My guess is it would be a natural logarithm. |
|
|
|
Post by besoeker3 on Nov 24, 2022 18:12:50 GMT
What logarithmic base do the neurons use? He does not say and the paper is unavailable. My guess is it would be a natural logarithm. Or Napierian logarithms. or Napier's bones. |
|
|
|
Post by Toreador on Nov 24, 2022 18:46:35 GMT
He does not say and the paper is unavailable. My guess is it would be a natural logarithm. Or Napierian logarithms. or Napier's bones. Maybe even anti-logarithms. |
|
|
|
Post by colbops on Nov 24, 2022 21:32:39 GMT
Or Napierian logarithms. or Napier's bones. Maybe even anti-logarithms. Sounds like gibberisms to me |
|
|
|
Post by besoeker3 on Nov 24, 2022 22:30:59 GMT
Maybe even anti-logarithms. Sounds like gibberisms to me You you who John Napier Was? |
|
|
|
Post by colbops on Nov 24, 2022 23:06:48 GMT
Sounds like gibberisms to me You you who John Napier Was? He does not say and the paper is unavailable. My guess is it would be some guy that used to go round in circles playing with triangles |
|
|
|
Post by borchester on Nov 24, 2022 23:34:25 GMT
You you who John Napier Was? He does not say and the paper is unavailable. My guess is it would be some guy that used to go round in circles playing with triangles Napier was the chap who was most famous for his meeting with Henry Briggs in 1616, during which he cried
"Dog's bollocks Harry, logs to the base 10! Would you ever Adam and Eve it !"
Later on Napier went on to invent Cockney rhyming slang and thus became known as the father of the TV series Minder
|
|
|
|
Post by besoeker3 on Nov 25, 2022 15:44:42 GMT
Maybe even anti-logarithms. Sounds like gibberisms to me John Napier of Merchiston, nicknamed Marvellous Merchiston, was a Scottish landowner known as a mathematician, physicist, and astronomer. He was the 8th Laird of Merchiston. His Latinized name was Ioannes Neper. John Napier is best known as the discoverer of logarithms. Napierian logarithms. Merchiston Castle in Edinburgh is where is where I got my degree. |
|
|
|
Post by Toreador on Nov 25, 2022 16:26:08 GMT
Sounds like gibberisms to me John Napier of Merchiston, nicknamed Marvellous Merchiston, was a Scottish landowner known as a mathematician, physicist, and astronomer. He was the 8th Laird of Merchiston. His Latinized name was Ioannes Neper. John Napier is best known as the discoverer of logarithms. Napierian logarithms. Merchiston Castle in Edinburgh is where is where I got my degree. Does anyone use logarithms these days? |
|
|
|
Post by colbops on Nov 25, 2022 18:56:10 GMT
John Napier of Merchiston, nicknamed Marvellous Merchiston, was a Scottish landowner known as a mathematician, physicist, and astronomer. He was the 8th Laird of Merchiston. His Latinized name was Ioannes Neper. John Napier is best known as the discoverer of logarithms. Napierian logarithms. Merchiston Castle in Edinburgh is where is where I got my degree. Does anyone use logarithms these days? Sure. It usually does the trick when I've got a clinger situation. Doesn't happen as much these days granted, as my diet is much better than it was back in the day. |
|
|
|
Post by besoeker3 on Nov 25, 2022 18:58:52 GMT
John Napier of Merchiston, nicknamed Marvellous Merchiston, was a Scottish landowner known as a mathematician, physicist, and astronomer. He was the 8th Laird of Merchiston. His Latinized name was Ioannes Neper. John Napier is best known as the discoverer of logarithms. Napierian logarithms. Merchiston Castle in Edinburgh is where is where I got my degree. Does anyone use logarithms these days? Yes, Naperian Logathirms and the exponential number e. |
|
