SRMTLShree Rama Multi-Tech Limited (India)
Copyright 1988-2018, All rights reserved.
References in periodicals archive ?
(7) Sparse regularized multitask learning formulation (SRMTL) [26]: SRMTL (([min.sub.[THETA]] L(X,Y,[THETA]) + [[lambda].sub.1][[parallel][THETA]L[parallel].sup.2.sub.F] + [[lambda].sub.2][[parallel][THETA][parallel].sub.1], where L [member of] [R.sup.TxT]) containing two regularization processes: (1) all tasks are regularized by their mean value, and therefore knowledge from one task can be utilized by other tasks via the mean value; (2) sparsity is enforced in the learning with [l.sub.1]-norm.