Robust Dendritic Computations With Sparse Distributed Representations

Click on the image above to enlarge

Cortex encodes information with sparse distributed patterns. How accurately can neurons recognize sparse patterns?

This poster is very similar to a poster we presented at Cosyne 2018 in March on the same topic of sparse distributed representations. It examines how accurately neurons can recognize these sparse patterns by showing:

  1. Scaling laws for computing error probabilities.
  2. High dimensional sparse patterns can be classified extremely reliably, even with large amounts of noise.
  3. Active dendritic segments can reliably classify patterns using a tiny number of synapses.
  4. The equations explain experimentally observed NMDA spike thresholds in active dendrites.
  5. Behavior of Poirazi-Mel and HTM neuron models closely match theoretical predictions. Understanding the behavior can lead to dramatically improved accuracies