In the field of machine learning (ML), locally linear embedding (LLE) algorithm, as a nonlinear dimensionality reduction method of manifold learning, has become an indispensable tool, is very effective in handling high-dimensional data arrays. The core idea of LLE algorithm is to find the proper mapping of data points from high-dimensional space to low-dimensional space, with their algebraic topology unchanged. LLE is usually transformed into two optimization problems, constrained optimization and unconstrained optimization, which are usually solved by gradient descent (GD) method or stochastic gradient descent (SGD) method. However, gradient dependent methods often fall into the trap of local minimum and cannot find the global optimal solution. Based on the idea of employing swarm intelligence algorithm, in this paper, a t-distribution sailfish optimization algorithm has been introduced to overcome the problem of global optimization in the original LLE method. Via numerical experiment, it has been proved that the proposed sailfish optimization algorithm can improve the robustness and the effectiveness of the algorithm.