Abstract
Many different approaches to “low-level” perception and action processing are possible within the overall CogPrime framework. We discussed several in the previous chapter, all elaborations of the general hierarchical pattern recognition approach.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
While the Numenta system is the best-known CSDLN architecture, other CSDLNs appear more impressively functional in various respects; and many CSDLN-related ideas existed in the literature well before Numenta’s advent.
- 2.
- 3.
- 4.
The perceptual CSDLN shown is unrealistically small for complex vision processing (only four layers), and only a fragment of the semantic-perceptual CSDLN is shown (a node corresponding to the category face, and then a child network containing nodes corresponding to several components of a typical face). In a real semantic-perceptual CSDLN, there would be many other nodes on the same level as the face node, many other parts to the face subnetwork besides the eyes, nose and mouth depicted here; the eye, nose and mouth nodes would also have child subnetworks; there would be link from each semantic node to centroids within a large number of perceptual nodes; and there would also be many nodes not corresponding clearly to any single English language concept like eye, nose, face, etc.
- 5.
In the figure, only a fragment of the semantic-motoric CSDLN is shown (a node corresponding to the “get object” action category, and then a child network containing nodes corresponding to several components of the action). In a real semantic-motoric CSDLN, there would be many other nodes on the same level as the get-object node, many other parts to the get-object subnetwork besides the ones depicted here; the subnetwork nodes would also have child subnetworks; there would be link from each semantic node to centroids within a large number of motoric nodes; and there might also be many nodes not corresponding clearly to any single English language concept like “grasp object” etc.
- 6.
The diagram is simplified in many ways, e.g. only a handful of nodes in each hierarchy is shown (rather than the whole hierarchy), and lines without arrows are used to indicate bidirectional arrows, and nearly all links are omitted. The purpose is just to show the general character of interaction between the components in a simplified context.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
Copyright information
© 2014 Atlantis Press and the authors
About this chapter
Cite this chapter
Goertzel, B., Pennachin, C., Geisweiller, N. (2014). Integrating CogPrime with a Compositional Spatiotemporal Deep Learning Network. In: Engineering General Intelligence, Part 2. Atlantis Thinking Machines, vol 6. Atlantis Press, Paris. https://doi.org/10.2991/978-94-6239-030-0_9
Download citation
DOI: https://doi.org/10.2991/978-94-6239-030-0_9
Published:
Publisher Name: Atlantis Press, Paris
Print ISBN: 978-94-6239-029-4
Online ISBN: 978-94-6239-030-0
eBook Packages: Computer ScienceComputer Science (R0)