Numenta Logo

A Neural Mechanism for Sequence Learning – HTM Sequence Memory

Bernstein Conference 2016: HTM Sequence Memory for Sequence Learning

Click on the image above to enlarge

Sequence learning is ubiquitous in cortex. This poster explains a neural mechanism for sequence learning called HTM (Hierarchical Temporal Memory) Sequence Memory.

Characteristics of HTM sequence memory:

  1. Neurons learn to recognize hundreds of patterns using active dendrites.
  2. Recognition of patterns act as predictions by depolarizing the cell without generating an immediate action potential.
  3. A network of neurons with active dendrites forms a powerful sequence memory.
  4. Sparse representations lead to highly robust recognition.

HTM sequence memory uses highly sparse representations, which makes it highly fault tolerant and robust to noise. Because it adapts quickly to changes due to its ability to learn continuously, it works well on real-world data.