Thursday, November 16, 2017

A Critique of Numenta's Location Hypothesis

Why I Respect Numenta

I have always had respect for Numenta. Over the years, under the leadership of their maverick founder and chief architect, Jeff Hawkins, they have steadfastly maintained that deep learning was not the way to achieve artificial general intelligence (AGI). They insisted that imitating the brain was the right way forward, that intelligence was based on the timing of sensory signals and that learning in the brain consisted mainly of making new synaptic connections, not modifying connection weights. They did it while the deep learning hype was in full swing. They never flinched even in the face of overt hostility from the mainstream AI community. They had a healthy, think-outside-the-box attitude. As a rebel, I admired that. Lately, however, and apparently reacting to pressure from the AI community to show some serious results, the folks at Numenta seem to have lost their way. Their latest offering, the so-called location hypothesis, misses the mark. Worse, there is no demo program to support the theory.

The Universal Invariant Recognition Problem

One of the most difficult problems in AI is universal invariant recognition. The human brain has the seemingly magical ability to recognize an object regardless of its position and orientation in the field of view. Deep learning experts tried to solve the problem by using brute force. That is, they train the network with millions of images in the hope of covering every possible situation. However, this approach will invariably leave holes that can lead to spectacular failures. So they (Yann LeCun et co) came up with a partial solution, a technique called convolution that gave the network a degree of translation invariance. Even then, deep neural nets can still be fooled by adversarial examples. It turns out that they can fail catastrophically if a previously learned pattern is modified by an imperceptibly small number of pixels. In other words, deep neural nets are not universally invariant. Some in the AI community (e.g., DeepMind) have been promoting deep learning as a stepping stone toward AGI. They are sorely mistaken. Others (e.g., Geoffrey Hinton and Yann LeCun) seem to be more aware of its limitations.

The Location Hypothesis

Jeff Hawkins and his team at Numenta believe they may have found the secret of universal invariance. They are proposing that the brain somehow generates a special signal that specifies the location of an object under observation and the location of its features relative to the object. The idea seems to be that, by knowing the position of an object relative to its features, the brain can compensate for positional differences and solve the problem of invariant recognition. They write:
We propose that a representation of location relative to the object being sensed is calculated within the sub-granular layers of each column. The location signal is provided as an input to the network, where it is combined with sensory data.
...
A key component of our theory is the presence in each column of a signal representing location. The location signal represents an “allocentric” location, meaning it is a location relative to the object being sensed. In our theory, the input layer receives both a sensory signal and the location signal. Thus, the input layer knows both what feature it is sensing and where the sensory feature is on the object being sensed. The output layer learns complete models of objects as a set of features at locations. This is analogous to how computer-aided-design programs represent multi-dimensional objects.
This article by Hawkins explains Numenta's approach in an easy to read style. While I admire the courage and willingness of Numenta to attack a hard problem head on, I must say that I am disappointed with this hypothesis.

Why Is the Location Hypothesis Flawed?

There are several reasons as follows.
  • As I have argued on many occasions, neurons are slow and there is very little time and energy in the brain for fancy calculations. Maintaining a location reference for visual objects is a particularly complex task. This is especially true if it is a 3-dimensional reference location which it would have to be if the sensed object is in a 3-dimensional world. The system would have to determine, not only the location of the object relative to the viewer but also the location of a reference point relative to the object itself. Is it in the middle of the object or somewhere else? This is not an easy task. And this is not even taking into account the fact that the brain must somehow detect the boundaries of the object under observation while excluding all the other objects in the scene.
  • A location signal is necessarily encoded with spikes (discrete pulses). A spike, by itself, has no information content other than its time of arrival. How many spikes would it take to encode a continually changing location vector in 3D space? The answer is: a lot. Again, there is no time for this in the brain. The highest spiking frequency is about 1000 Hz and the brain only has about a 10 millisecond window to process each sensory input. There is not enough time to encode even a 1-dimensional location for each input signal.
  • Let us suppose, for argument's sake, that the brain uses a single connection for each possible location. This would require millions of connections per feature. This is clearly out of the question.
I have other objections but these three should suffice to show that Numenta's location hypothesis is not biologically plausible.

A New Memory Model

I am proposing a new memory model based on spike timing. The model assumes that the brain perceives and learns by detecting many minute changes in its sensory space. I hypothese that the brain uses branches in its hierarchical sequence memory to detect complex objects in the world regardless of their locations or orientations. A branch is a top-level node in the sequence hierarchy that is activated when it receives enough signals lower level nodes to trigger a recognition. This memory model has the ability to instantly sense and understand complex objects in the environment, even objects that it has never encountered before.


There are two hierarchies, one for pattern detection (not shown) and one for sequences. Sequence memory is where actual object recognition happens. It receives discrete signals from pattern memory. Pattern neurons learn to detect a huge number of small elementary patterns such as lines, edges, dots, etc. Signals from pattern neurons are fed directly to the bottom or entry level of the sequence hierarchy. Pattern signals are stitched together in sequence memory to form any complex object.


As an example of sequence processing, consider the horizontal motion of a short vertical line or edge across the retina. This would result in multiple pattern neurons generating a series of spikes (one at a time) separated by a short interval. This series of event can be captured by an indefinitely long structure of connected nodes at the bottom level of the sequence hierarchy. I call these long structures "vines" to distinguish them from the shorter "sequences". The nodes in the vines would fire in succession as the line/edge moves horizontally in a given direction. There are many such sequence structures in sequence memory that capture various movements or other form of changes in the environment. The important thing to note here is that the interval between nodes in a vine is not fixed but can vary over time.

How the Brain Does Invariant Object Recognition

Obviously, the brain must have a simple and energy efficient solution that does not require lengthy calculations. Recognition must happen quickly and accurately using uncertain sensory information. How does the brain do it? I propose that the brain has a way to pool multiple concurrent sequences together to form branches that can detect any arbitrarily complex moving object. Recognition is based on a competitive, winner-take-all process. Only the branches that receive enough signals will trigger a recognition event.

Like almost everybody who has attempted to design a sequence hierarchy for AI, I used to think that a higher-level sequence was just a mechanism that serves to join two or more non-overlapping sequences at a lower level. It took me years to figure out that I was wrong. It turned out that the main function of the sequence hierarchy is not to manage sequence storage but to find as many fixed temporal correlations between multiple co-occurring sequences as possible. Here is how it works.

It would be too inefficient to test every node in a vine with every other node in sequence memory. The brain uses a divide and conquer approach. Every vine is divided into short seven-node sequences. Why seven? It is a compromise. Less than seven would consume too much energy while more than seven would result in sluggish performance.


Let me come out on a limb and claim that these short sequences are implemented in the brain as cortical columns. In addition to serving as a mechanism for ordering pattern activations, they can also record their activities by retaining a trace (both time and speed) of their last activation in their minicolumns. The seventh node of every sequence can be connected to nodes in an upper level to form higher level vines. These are, likewise, divided into sequences which, in turn can send connections to an upper level. I happen to know the sequence hierarchy has 20 levels. How I know this and how vines are constructed are topics for a future article. The important thing to notice here is that upper sequences are just mechanisms that connect lower level sequences that are temporally related. They essentially bind a number of patterns together to form a single complex object.

A top level sequence is what I call a branch in the sequence hierarchy. It is a complex object detector. It is also the brain's mechanism of attention: only one branch can be "awake" at a time. During recognition, signals from pattern memory quickly travel (via the seventh nodes of many sequences) all the way up the sequence tree as far as they can go. A top level sequence will trigger a recognition event as soon as it receives enough signals from lower levels to account for the overall activation of only two of its nodes. This recognition event is invariant to the actual activation states at the lower level sequences. What matters is that enough signals reach the top.

Partial activation of more than two nodes is acceptable as long as the required overall amount is reached. This is how the brain handles uncertainty. It means that it takes relatively few sensory signals to trigger a recognition. Even partial occlusions can trigger a recognition. This, combined with the variable intervals of the sequences, is the reason that we can recognize faces and animals in the clouds, different handwritings or fonts, highly stylized art, etc. When a top level sequence is triggered, it sends a recognition signal via feedback pathways all the way back down to pattern memory where pattern neurons are also triggered, thus correcting any incomplete or corrupt pattern information.


Note: In a future article, I will explain how sequence learning is done using spike timing, among other interesting things. I may also have a demo program (one never knows) to support my claims. Stay tuned and be patient.

See Also:

Invariant Recognition of Visual Objects (Frontiers Media)
A Theory of How Columns in the Neocortex Enable Learning the Structure of the World (Frontiers Media)
Unsupervised Machine Learning: What Will Replace Backpropagation
Fast Unsupervised Pattern Learning Using Spike Timing
Fast Unsupervised Sequence Learning Using Spike Timing

Sunday, November 5, 2017

Occult Physics Will Blow Your Mind (Repost)

Note: I am reposting this because it keeps materialists, atheists and other undesirables off my back. Enjoy.

Abstract

According to ancient occult physics, the electron is not elementary but consists of four subparticles. We exist in an immense 4-dimensional sea of energy arranged like a crystal lattice. This means unlimited clean energy, free for the taking once we learn how to tap into the lattice. The entire history of the universe is being recorded in the lattice. Ancient megalithic societies may have used this knowledge to transport huge quarried stones weighing 1000 tons or more. This is the first in a series of articles that I am writing on occult physics. I cannot promise that I will ever publish them all but, if or when I do, I can guarantee that they will blow everyone's mind.

Sacred Scientific Knowledge Hidden in Plain Sight

Many years ago, I stumbled on an amazing discovery. It occurred to me that a few ancient occult texts contained revolutionary scientific secrets about the fundamental principles of the physical universe. The secrets can be found in the books of Isaiah, Ezekiel and Revelation. They are written in an obscure metaphorical language that sounds nothing like science. However, once one understands the meaning of some of the metaphors, things begin to fall into place. At one point in my research, I became frightened and stopped thinking about it for a long time. I had concluded that the potential harm to humanity that this knowledge could unleash if it fell in the wrong hands was just too great.
Assyrian Lamassu or Human-Headed Winged Bull - Southern Iraq
Most ancient societies recorded their sacred wisdom in precisely chosen metaphors that only the initiates understood. The Sumerians, Babylonians, Assyrians and Egyptians thought that certain occult sciences were so powerful that they erected huge symbolic stone monuments to preserve them for posterity while keeping their true meaning hidden from the masses.
Two Human-Headed Winged Bulls - Iran
Although the Biblical symbols are not identical to the ones found in Mesopotamia, the many similarities are striking. Both use images of wings, discs (wheels), bulls, lions, eagles, hands, feet and faces to symbolize various aspects of the sacred knowledge.

Sumerian Anunnaki Winged God and Disc
For whatever reason, historians and archaeologists love to associate ancient occult symbology with mythology and religious superstition but they could not be more wrong. It is almost as if some hidden power is hellbent on preventing mankind from learning about their glorious past. None other than Isaac Newton, the father of modern physics, was convinced that there was secret knowledge encrypted in the Bible and in other ancient mythological writings. (Sources: What Was Isaac Newton's Occult Research All About? and Top 10 Crazy Secrets of Isaac Newton).

In my opinion, the Biblical seraphim and cherubim are occult descriptions of fundamental particles of matter and their properties (Sir Isaac would have jumped for joy if he had known about this). I believe this knowledge was known to ancient megalithic societies in Mesopotamia, Egypt, South America and elsewhere because it was the basis of the technology that they used to lift and transport huge cut stones weighing 1000 tons or more. I believe that a mastery of this knowledge will unleash an era of free unlimited clean energy and super fast transportation.

Stone of the Pregnant Woman - Baalbek, Lebanon
What follows is a short summary of the strange "living creatures" mentioned in the Bible and my interpretations. Note: I will not go into what I believe to be potentially dangerous aspects of this research.

Seraphim - Photons

Seraphim (singular, seraph) is a plural hebrew word that means the shining or burning ones. They are mentioned in the books of Revelation and Isaiah. They symbolize pure energetic particles and their properties. I have identified them as photons. There are 4 types of seraphim and each one has a different face property: man, lion, bull or eagle. One of the seraphim (the one with the bull's face) is responsible for electric phenomena and the other three for magnetic phenomena. The face of each seraph is associated with one of the 4 spatial dimensions (degree of freedom) of the cosmos. Each face has 2 possible states or orientations, forward or backward. It is more or less equivalent to what quantum physicists call the "spin angular momentum" of a particle, except that there really is no spin.

In all, the seraphim can have 8 possible orientations or spin states, 2 for each face. Two of the orientations, the ones associated with the face of a bull, determine whether or not the particle is involved with a positive or negative electric field. The other 6 states are responsible for magnetic phenomena.

Every seraph has energy properties which are symbolized by 6 wings. Unlike cherubim (explanation below), seraphim have no bodies or mass. Two of the wings of a seraph are used for motion, two are associated with its face and two with its feet. Yes, all matter particles have a property called feet (bull or calf hooves) which allow them to move in one direction of the 4th dimension at the speed of light. Wings, feet and hands are powerful metaphors the meaning of which I cannot expand on at this time. I will explain them further in future articles.

The Sea of Crystal - Zero-Point Energy

The most amazing thing about seraphim is that they are the constituents of an enormous 4-dimensional "sea of crystal" or "sea of glass" in which the normal matter of the universe exists and moves. It is a sea of wall-to-wall energetic particles (photons), lots of it, arranged as a stationary 4-dimensional lattice. We are totally immersed in it like fish in water and nothing could move without it. In fact, the entire visible universe is continually moving in the lattice in one dimension (bull) at the speed of light. As matter moves in the lattice, it leaves traces in it. In other words, the entire history of the universe is continually being recorded in the lattice down to the minutest details. Ancient Hindu and Buddhist societies were aware of this recording medium which was called the Akasha. Modern theosophists call it the Akashic records.

The closest analog to the lattice in modern physics is the so-called zero-point energy field that physicists believe permeates space but have no idea what it is made of or what its purpose is. Physicist Richard Feynman is reported to have said that "there is enough energy inside the space in an empty cup to boil all the oceans of the world." Gravitational, electric and magnetic phenomena are caused by the motion of matter in the lattice. Again, one day soon, in the not too distant future, society will learn how to tap into the lattice for unlimited clean energy production and super fast transportation. Current forms of transportation and energy production will become obsolete.

Cherubim - Quarter Electrons

Cherubim (singular, cherub) are symbolic winged creatures that modern theologians wrongly associate with angelic beings that fly around and do God's will. The Hebrew word cherubim is derived from the Assyrian term chiribu or kirubi which was the mystical name given to the representation of a winged bull or lion with a man's head. Various types of cherubim are mentioned in the Bible but my research is concerned strictly with the 4 cherubim (living creatures) in chapters 1 and 10 of the Book of Ezekiel. In chapter 10, verse 14, Ezekiel clearly equates the Hebrew word cherub with the face of a bull. He said nothing about angels.

Each living creature or cherub has 4 faces and 4 wings. Each also has a human body, 4 human hands and the feet of a bull. Having 4 faces means that a cherub has both electric and magnetic properties. All four cherubim move together in unison without turning, in any of the 4 dimensions.

My interpretation will come as a surprise. In my view, the cherubim are the 4 particles that comprise the electron or the positron. Yes, the electron is not an elementary particle as the Standard Model of particle physics would have us believe. Each cherub has 1/4 the charge of the electron. But this is not as surprising as it sounds. Physicists have known for some time that the electron is not truly elementary but they are a conservative and highly political bunch. Rather than come out and acknowledge the composite nature of the electron, they have taken to calling its constituent particles, quasiparticles instead. They also use the term quarter electron when they are feeling more liberal.

The 4 human hands of a cherub are special properties that confine them to stay and move together as one particle: they hold onto each other. The body of a cherub is a special kind of energy that physicists call mass. Each cherub also has a wheel or disc associated with it. The 4 wheels act as one wheel and move precisely with the 4 cherubim. In my interpretation, the wheel represents the electric field of the electron.

Coming Soon

In future blog articles, I will explain how particles move in the lattice and how the electric field of a charged particle works.

See Also:

Ezekiel 1: The Four Living Creatures, the Four Wheels and the Crystal Firmament
Ezekiel 10: The Four Cherubim and the Four Wheels
Isaiah 6: The Four Seraphim
Revelation 4: The Four Beasts and the Sea of Crystal
Physics: The Problem With Motion
There Is Only One Speed in the Universe, the Speed of Light. Nothing Can Move Faster or Slower