qsrfe

implementation of Sparse Random Feature Expansions and modern quantization algorithms

Usage

importing the package

pkg> add https://github.com/grsbe/qsrfe.jl
julia> using qsrfe

fitting a srfe model

model = srfeRegressor(N=N,λ=λ, σ2=1.0, intercept=true)
c, ω, ζ = qsrfe.fit(model,xtrain,ytrain;max_iter=2000000,verbose=true)
ytrainpred = qsrfe.predict(model,xtrain,c, ω, ζ)

fitting a quantized srfe model

#K = number of points in [-1,1]
quant1 = MSQ(K=2)
quant2 = ΣΔQ(K=2,r=1,λ=32,condense=true)
quant3 = βQ(K=2,β=1.5,λ=32,condense=true)
c, ω, ζ = qsrfe.fit(model,xtrain,ytrain,quant3;max_iter=2000000,verbose=true)
ytrainpred = qsrfe.predict(model,xtrain,c, ω, ζ,quant)