Found an error in SDR paper union false probability calculation

While coding up more SDR visualizations, I found an error in Properties of Sparse Distributed Representations and their Application to Hierarchical Temporal Memory.

Formula (13) in section G. The Surprising Union Property is incorrect:

The right side of the equation should read:

The newer and more neuroscience-y version of the SDR paper at How do neurons operate on sparse distributed representations? A mathematical theory of sparsity, neurons and active dendrites has the proper calculation.

1 Like