### Saturday, April 04, 2015

## Compression and representation

In a maximally lossless compressed representation all relationships are preserved and the loss of any bit will result in the loss of information. Therefore the evolution of a representation can be analyzed in terms of lossy and noloss compression. If there is a compression to a region of a space, this can be described in terms of minimizing entropy.

(0) comments
You want to reduce the size of semantic voids and differentiate any aliased states. Continuity can be one kind of relationship where you constrain the number of directions a membrane diffuses. A point in model space should be close to the average of its neighbors in input space.

Topics: Representation | compression

### Friday, March 06, 2015

## Boundaries and prototypes

Prototypes should trace the edges of a boundary, with positive errors inside and negative errors outside. Balancing to zero

(0) comments
Topics: representation | neural networks

## Dimensional transformations

Not only must input prototypes be cultivated to support a compressed representation, but multiple outputs can be generated to provide multiple sources of error from different overlapping lower order projections.

(0) comments
Topics: neuralnetworks | representation

### Saturday, February 21, 2015

## Fractal dimensionality as a requirement for learning

Under what conditions is self organization possible? What are the constrains on the search space that are needed for exponential rates of increased learning?

While learning may be impossible in an exponentially large multidimensional space, learning may be possible if the interesting points in the space can be compressed to a more compact volume of the space as a starting point for search.

Can the small number of "legal/coherent/valued" points within the space be compressed to a relatively compact fractal volume within a vast expanse of valueless space.

If one of the dimensions used to encode this compressed space was entropy would movement along that dimension appear as time.

Is this pattern observed in the evolution of life?

(0) comments
While learning may be impossible in an exponentially large multidimensional space, learning may be possible if the interesting points in the space can be compressed to a more compact volume of the space as a starting point for search.

Can the small number of "legal/coherent/valued" points within the space be compressed to a relatively compact fractal volume within a vast expanse of valueless space.

If one of the dimensions used to encode this compressed space was entropy would movement along that dimension appear as time.

Is this pattern observed in the evolution of life?

Topics: Self Organization | Morphogenesis

### Monday, June 02, 2014

## Conservation laws and self-organization

perturb

compensate

overshoot

compensate

overshoot more (feed-forward)

. . .

reach limit

compensate

overshoot less (feed-back)

compensate

balance

centrifugal/evolution

centripetal/involution

relationship to symmetry and conservation laws using 3D editing as a metaphor:

Shift 3D scene to allow 2D symmetry enhancement from a limited view

If the alignment is wrong then the local symmetry can destroy global symmetry

If the alignment is close enough then the problem can be overcome with a series of small local adjustments from different 2D views

2D Symmetry around a point involves summing to zero (conservation in one dimension)

2D symmetry around a line involves mapping pairs of point across an axis that sums to zero

The point or axis of symmetry provides a kind of compression function

(0) comments
compensate

overshoot

compensate

overshoot more (feed-forward)

. . .

reach limit

compensate

overshoot less (feed-back)

compensate

balance

centrifugal/evolution

centripetal/involution

relationship to symmetry and conservation laws using 3D editing as a metaphor:

Shift 3D scene to allow 2D symmetry enhancement from a limited view

If the alignment is wrong then the local symmetry can destroy global symmetry

If the alignment is close enough then the problem can be overcome with a series of small local adjustments from different 2D views

2D Symmetry around a point involves summing to zero (conservation in one dimension)

2D symmetry around a line involves mapping pairs of point across an axis that sums to zero

The point or axis of symmetry provides a kind of compression function

### Saturday, May 31, 2014

## Anatomy Of A Dance Hit

Anatomy Of A Dance Hit

Incomplete Nature: How Mind Emerged from Matter

(0) comments
Last month neuroscientists at Aarhus University in Denmark a study showing thatThis strangely reminds me of one of the major themes of Terrance Deacon's resent work.danceable grooves have just the right amount of gaps or breaks in the beats.Your brain wants to fill in those gaps with body movement, says the study's lead author, .

"Gaps in the rhythmic structure, gaps in the sort of underlying beat of the music — that sort of provides us with an opportunity to physically inhabit those gaps and fill in those gaps with our own bodies," she says.

Incomplete Nature: How Mind Emerged from Matter

Deacon makes the case for absence as the critical factor in the transition from inorganic matter to life and from life to mind. To best illustrate this notion, ideas have no physical attribute yet they have casual power.See

The development of life and consciousness took the world literally from brute ‘matter to mattering. In this process nothing, according to Deacon, was added to (or subtracted from) ‘the matter, energy, or fundamental physical laws of nature’.

What was novel were the new forms of relational organization of the material and energetic dimensions of the universe inconstrained dynamical processes, and constraints are absences conserved as information within bioforms; and these new forms of organization and relation produced a new irreducible kind of cosmic causality with the emergence of ententionality.

Topics: |

### Monday, March 17, 2014

## Meaning and Destiny

If you pour water on the ground and it spells out your name, the information is not in the water, but in the contours of the surface of the ground. Quantum fields can be thought of as the probabilistic contours of superposition, or attractors, that define the information of evolving life. We trace the contours of our mind in order to reveal the pattern of our life. Any deviation, through resistance or ignorance, from this destiny creates a dam or potential of momentum whose contours will have to be learned and retraced in another cycle of time.

We are not able to change the set contours of our destiny, but we can change how we react to it, and create or dissolve, wind or unwind, the field of possibilities that encode the future.

### Sunday, February 17, 2013

## Innovation as evolution on the edge of dimensional chaos

Mapping the product/market space – an hallucination

(0) comments

Imagine a vast multi-dimensional space. Each point of it represents a specific need that a specific person has, an iota of utility.

1) A bad idea for a product business pops into quantum reality and immediately gets sucked into the huge black hole of a nearby product that is very flexible.

2) Much later, a second person (maybe after talking in the pub to the sister of somebody who years previously read a tweet by the first person) has almost exactly but not quite the same idea. They make a prototype

3) Finally, a third person realises something important about the market of the new product idea and pushes it along the space to a fatter area. Anyway, time has passed and all the dimensions have reconfigured somewhat. Finally, a new glowing inferno burns in the gap.

A startup is by definition a hunt for one of those spaces, or a subspace within one, just the right size and place for the team, and the time.