ABSTRACT
The previous chapters presented the fundamentals as well as an overview of multilinear subspace learning (MSL). In this final chapter of Part I, we discuss the algorithmic and computational aspects of MSL. This chapter is mainly for those who are interested in developing, implementing, and testing MSL algorithms. We first examine a typical iterative solution, as shown in Figure 5.1. We then discuss various issues, including initialization, projection order, termination, and convergence. Next, we consider the generation of a synthetic tensor dataset with varying characteristics for MSL algorithm analysis. Finally, we deal with feature selection strategies and various computational aspects.
