tsne_task: tSNE problem

In bhtang127/AccelBenchmark: Benchmark Various Accelerating Methods for Fixed Point Iterations

tSNE problem

Description

A function that will generate the task list that can be used to benchmark the acceleration of tSNE in COIL-20 data

Usage

tsne_task(perplexity = 30, rho = 2e-07, sd = 0.01)

Arguments

perplexity	Perplexity value used to calculate the similarity matrix in original space before embedding
rho	Initial Learning rate used in the MM algorithm for tSNE
sd	Standard deviation for the random initial embedding coordinates

Value

A list containing all components needed for benchmarking the problem

initfn	Parameter random initializing function
fixptfn	Updating function for the fixed point iteration problem
objfn	Function calculating the objective value for current parameter
...	Other arguments required in functions above

References

Yang Z, Peltonen J, Kaski S (2015). Majorization-minimization for manifold embedding. In: Artificial Intelligence and Statistics, 1088–1097. PMLR.

Examples

## Not run: 
set.seed(54321)
problem = tsne_task(rho=5e-6, sd=5e-3)
benchmark(
  problem,
  algorithm=c("raw", "squarem", "daarem", "pem", "qn", "nes"),
  ntimes = 20
)

## End(Not run)

bhtang127/AccelBenchmark documentation built on May 30, 2022, 2:21 a.m.