1 / 29

Networks of Neurons

Networks of Neurons. Computational Cognitive Neuroscience Randall O’Reilly. Networks. Biology of Neocortex (“cortex”) Categorization and Distributed Reps Bidirectional Excitation and Attractors Inhibitory Competition and Activity Regulation. Neurons: Excitatory and Inhibitory.

dung
Download Presentation

Networks of Neurons

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Networks of Neurons Computational Cognitive Neuroscience Randall O’Reilly

  2. Networks • Biology of Neocortex (“cortex”) • Categorization and Distributed Reps • Bidirectional Excitation and Attractors • Inhibitory Competition and Activity Regulation

  3. Neurons: Excitatory and Inhibitory Excitatory = main info processing, long-range connections Inhibitory = local, activity regulation and competition

  4. The 6 Layer Cake..

  5. Is Actually Only 3..

  6. Patterns of Connectivity

  7. Bidirectional Symmetry

  8. Biology => Function • Feedforward excitation = categorization of inputs • Feedback excitation = attractor dynamics • Lateral inhibition = competition, activity regulation

  9. We Think in Categories (much easier than disconnected pixels..)

  10. Hierarchy of Categories

  11. The Chair Category

  12. Getting the right ones is key.. • Two men are dead in a cabin in the woods. What happened??

  13. Categories are Interesting! • What makes a mental categorization accurate? Is there something “real” about a “chair?” • Stereotypes are mental categories.. • Can you encode multiple categories at the same time??

  14. Distributed Representations • Let a 1,000 categories bloom.. You’ve got the room in your head (billions of neurons) • Each neuron can respond to multiple things (graded similarity) • And each thing activates many neurons (who knows what is going to be relevant this time?)

  15. Graded Responses

  16. Distributed Patterns

  17. Topographic Organization

  18. Distributed Parts

  19. Not Just Monkeys

  20. Coarse Coding Efficiency

  21. Localist Representations?

  22. Bidirectional Excitatory Dynamics

  23. Top-down Ambiguity Resolution

  24. What Are These?

  25. A Big Network Model..

  26. Bidirectional Dynamics

  27. Inhibition • Competition: selection pressure, survival of the “fittest”, picking the best detector for the job.. • Interacts with learning: “rich get richer” (but also narrower – no hogging the inputs please!) • “Sparse distributed representations” • (and also essential for controlling activity, like an air conditioner)

  28. Feedforward and Feedback Inhib • Feedback “reacts” (AC comes on after it gets hot enough) • Feedforward “anticipates” (e.g.,if AC measured outdoor temp, or weather forecast)

  29. kWTA Approximation

More Related