Lagrangian Regularized Twin Extreme Learning Machine for Supervised and Semi-Supervised Classification

Twin extreme learning machine (TELM) is a phenomenon of symmetry that improves the performance traditional classification algorithm (ELM). Although TELM has been widely researched and applied in field learning, need to solve two quadratic programming problems (QPPs) for greatly limited its development. In this paper, we propose novel framework called Lagrangian regularized twin (LRTELM). One significant advantage our LRTELM over structural risk minimization principle implemented by introducing regularization term. Meanwhile, consider square l2-norm vector slack variables instead usual l1-norm order make objective functions strongly convex. Furthermore, simple fast iterative designed solving LRTELM, which only needs iteratively pair linear equations avoid QPPs. Last, extend semi-supervised manifold improve when insufficient labeled samples are available, as well obtain (Lap-LRTELM). Experimental results on most datasets show proposed Lap-LRTELM competitive terms accuracy efficiency compared state-of-the-art algorithms.