1. Classifying a Set of Vectors

Let \(X\) be a set of SDRs, and for any two SDRs \(x,y\) in \(X\), \[ match(x,y)=false, \text{if} x\neq y \] Then we can tell an SDR \(z\) belongs to \(X\) if \[ \exists x\in X, match(x,z) = true \] The idea behind set \(X\) is similar with nearest neighbour algorithm on the basis of some orthogonal vectors.

2. Union

Union is used to store a set of \(M\) vectors.

\(x_1 = 00100001\)

\(x_2 = 10001000\)

\(x_3 = 10000001\)

\(x_4 = 00101000\)

\(X = x_1 OR x_2 OR x_3 OR x_4 = 10101001\)

So that a fixed-size SDR vector can store a dynamic set of elements.

3. SDRs and HTM

HTM is a detailed computational theory of the neocortex.