Since that, many different regularization terms have been proposed.
With various constraints and regularization terms on the NMF objective function, there are many variants of NMF, which are appropriate for a large variety of problem domains.
Combining different regularization terms and various loss functions, we can have many variants of such linear models [ 2].
Appropriate regularization terms render the problem well-determined.
Besides, a meaningful regularization parameter is introduced for each problem in Lap-LSTSVM to balance the regularization terms between the reproducing kernel Hilbert spaces (RHKS) term and the manifold regularization (MR) term, instead of two parameters used in Lap-TSVM.
It consists of one error part and two regularization terms: the complexity term and the smoothness term.
We take into account the data specifics by incorporating multiple regularization terms into the optimization.
Its level set formulation combines length, region-based and regularization terms.
Here ha (a = 1, 2, 3) are first order spatial gradient filters for RGB channels, and ir is a sharp image in reference channel r, and α, β, γ and μ are weights for the regularization terms.
This leads to overall simple parameter setting of our method compared to many regularization methods.
Here, the divergence in the regularization terms is understood in the distributional meaning.
