solver.control: Auxiliary for controlling optimization routines

In evanwang1990/FMwR: Factorization Machines with R

Description

Auxiliary functions for optimization routines

Usage

solver.control(max_iter = 10000, convergence = 1e-4, evaluate.metric = "LL", solver = TDAP.solver())

Arguments

max_iter	integer giving maximal number of iterations, 25 for MCMC/ALS and 10000 for other optimization routines by default
solver	function to set parameters of optimization routine

Details

SGD/FTRL/TDAP updates model every single example, while MCMC/ALS needs to sweeps through all the data for each update. So the maximal number of iterations of MCMC/ALS is much smaller, which is 25 by default.

By now, only MCMC and ALS support parallel computing.

solver is a function to set parameters of optimization routine further, including ALS.solver MCMC.solver SGD.solver TDAP.solver FTRL.solver

evanwang1990/FMwR documentation built on May 16, 2019, 9:38 a.m.