Technique "Vector Quantization"

From SMW CindyKate by dataspects
Component0731215963
Jump to navigation Jump to search
[edit]
Keywords/Contexts

MachineLearning

Annotations
{{#compound_query:Component0731215963Property "-Has subobject" has a restricted application area and cannot be used as annotation property by a user.UseCaseMotivation;?HasDirection;?IsMotivating;?IsMotivatedBy;?HasReasoning
Property "-Has subobject" has a restricted application area and cannot be used as annotation property by a user.Technique "Vector Quantization";?HasDirection;?IsMotivating;?IsMotivatedBy;?HasReasoning Property "-Has subobject" has a restricted application area and cannot be used as annotation property by a user.Technique "Vector Quantization";?HasDirection;?IsMotivating;?IsMotivatedBy;?HasReasoning
name=MotivationsDeclaredOnThisPageResultItem|link=none}}
Motivations
Lua error in Module:Motivations at line 23: attempt to concatenate field 'hasReasoning' (a nil value).
[edit]

Content

Notice:

See Publication "How to create a mind", page 135.

Purpose

  1. Reduce data complexity
  2. Reduce multiple data dimensions to one (a single number as the quantized representation of a cluster)
  3. Improve ability to find invariants
    • Reduce data to equally likely probabilities
  4. Make it possible to use one-dimensional pattern recognizers

Steps

  1. How many clusters? E.g. 256.
  2. Register the first 256 one-point clusters.
  3. Take the 257th point and find its distance X to its closest neighbor (out of the already registered 256 points).
  4. If X is greater than the smallest distance of any pair of the already registered 256, then it is a new one-point cluster.
  5. Collapse the two one-point clusters that are closest together into a single cluster.
  6. Process remaining points while always maintaining 256 clusters.
  7. Find each cluster's geometric center point (vector).