Optimized product quantization (OPQ)

quantize_opq(X, R, C, V=false) -> B

Given data and PQ/OPQ codeboks, quantize.

Arguments

• X::Matrix{T}: d-by-n data to quantize
• R::Matrix{T}: d-by-d rotation to apply to the data before quantizing
• C::Vector{Matrix{T}}: m-long vector with d/m-by-h matrix entries. Each matrix is a (O)PQ codebook.
• V::Bool: Whether to print progress

Returns

• B::Matrix{Int16}: m-by-n matrix with the codes that approximate X

train_opq(X, m, h, niter, init, V=false) -> C, B, R, error

Trains an optimized product quantizer.

Arguments

• X::Matrix{T}: d-by-n data to quantize
• m::Integer: Number of codebooks
• h::Integer: Number of entries in each codebook (typically 256)
• niter::Integer: Number of iterations to use
• init::String: Method used to intiialize R, either "natural" (identity) or "random".
• V::Bool: Whether to print progress

Returns

• B::Matrix{Int16}: m-by-n matrix with the codes
• C::Vector{Matrix{T}}: m-long vector with d-by-h matrix entries. Each matrix is a codebook of size approximately d/m-by-h.
• R::Matrix{T}: d-by-d learned rotation for the data
• obj::Vector{T}: The quantization error each iteration