Inspiration of On-intelligence

Please note that all these ideas may prove to be wrong or will be revised.

Artificial Intelligence: wrong way

We are on the wrong way of Artificial General Intelligence(AGI).

The biggest mistake is the belief that intelligence is defined by intelligent behavior.
Object detection or other tasks are the manifestations of intelligence not the intelligence itself.

The great brain uses vast amounts of memory to create a model of the world, everything you know and have learned is stored in this model.

The ability to make predictions about the future that is the crux of intelligence.

Neural Networks:

  1. We must include time as brain function: real brains process rapidly changing streams of information.
  2. The importance of feedback: In thalamus(丘脑), connections going backward toward the input exceed the connections going forward by almost a factor of ten. But back propagation is not really feedback, because it is only occurred during the learning phase.
  3. Brain is organized as a repeating hierarchy.

History shows that the best solution to scitific problems are simple and elegant.

The Human Brain: all your knowledge of the world is a model based on patterns

  1. The neocortex is about 2 milimeters thick and has six layers, each approximated by one card.
  2. The mind is the creation of the cells in the brain. There is nothing else.And remember the cortex is built using a common repeated element.
  3. The cortex uses the same computational tool to accomplish everything it does.

According to Mountcastle‘s proposal:
The algorithm of cortex must be expressed independently of any particular function or sense.
The cortex does something universal that can be applied to any type of sensory or motor system.

When scientists and engineers try to understand vision or make computer that can “see”, they devise terminologies and techniques specific to vision.
They talk about edges, textures, and three-dimensional representations.
If they want to understand spoken language, they build algorithms based on rules of grammar, syntax and semantics.

But these approaches are not how the brain solves these problems, and are therefore likely to fail.

Attention mechanism:
About three times every second, your eyes make a sudden movement called a saccade.
Many vision research ignore saccades and the rapidly changing patterns of vision.

Existence may be objective, but the spatial-temporal pattern flowing into the axon bundles in our brains are all we have to go on.

Memory

The brain does not “compute” the answers to problems, it retrieves the answers from memory.
The entire cortex is a memory system rather than a computer at all.

The memory is invariant representations, which handle variations in the world automatically.

  • The neocortex stores sequences of patterns
    There are thousands of detailed memories stored in the synapses of our brains that are rarely used.
    At any point in time we recall only a tiny fraction of what we know.(remind A-Z is easy, Z-A is hard)

  • The neocortex recalls patterns auto-associatively.
    Your eyes only see parts of a body, but your brain fills in the rest.
    At any time, a piece can activate the whole. This is the essence of auto-associative memories or inferring.
    Thought and memories are associately linked, notice that random thoughts never really occur!

  • The neocortex stores patterns in an invariant form.
    We do not remember or recall things with complete fidelity.
    Because the brain remembers the important relationships in the world, independent of the details.
    To make a specific prediction, the brain must combine knowledge of the invariant structure with the most recent details.

    When listening to a familar song played on a piano, your cortex predicts the next note before it is played. And when listening to people speak, you often know what they are going to say before they have finished speaking.

  • The neocortex stores patterns in a hierarchy.

A New Framework of Intelligence: Hierarchy

The brain is using memories to form predictions about what it expects to experience before experience it.
When prediction is violated, attention is drawn to the error.
Incorret predictions result in confusion and prompt you to pay attention.
Your brain has made a model of the world and is constantly checking that model against reality.

By comparing the actual sensory input with recalled memory, the animal not only understands where it is but can see into the future.

How the Cortex Works

If you don’t have a picture of puzzle’s solution, the bottom-up method is sometimes the only way to proceed.

Here is an interesting metaphor:

Many of puzzle pieces will not be used in the ultimate solution, but you don’t know which one or how many.

I can not approve the ideas from Hawkins in this part. Still we don’t know how the cortex works actually.

How the Cortex Learns

Donlad O.Hebb, Hebbian learing: When two neurons fire at the same tiem, the synapses between them get strengthened

  1. Forming the classifications of patterns.
  2. Building memory sequences.

Note that prior to neocortex, the brain has:

  1. The Basal ganglia(基底神经节): Primitive motor system.
  2. The cerebellum(小脑): Leared precise timing relationships of evenets.
  3. The hippocampus(海马体): stored memories of specific events and places.

The hippocampus is the top region of the neocortex, not a separate structure.

There are many more secrets to be discovered than we currently know

Reference

请zzy824喝杯咖啡
0%