Hi Koobas, We have two classes that implement the solutions described in the ALS-related papers:
For explicit feedback data [1] we have AlternatingLeastSquaresSolver, for implicit feedback data [2] we have ImplicitFeedback-AlternatingLeastSquaresSolver. Both can be found in the org.apache.mahout.math.als package. Internally the use Mahout's QRDecomposition to solve the linear systems associated with ALS. Best, Sebastian [1] http://www.hpl.hp.com/personal/Robert_Schreiber/papers/2008%20AAIM%20Netflix/netflix_aaim08(submitted).pdf [2] http://research.yahoo.com/pub/2433 On 08.01.2013 21:53, Koobas wrote: > Perhaps somebody can shed some light on the topic. > What algorithm is used to solve the least squares problem > when computing low-rank approximation using Alternating Least Squares? >
