Abstract: This paper proposes a new efficient adaptive marine predators algorithm in light of the shortcomings of the marine predators algorithm in complex functions and engineering design optimization problems, including limited adaptive ability, low optimization accuracy, and high local shackle probability. Firstly, the learning automata-guided teaching-learning search mechanism is introduced in the marine memory stage, which better balances the exploration and exploitation requirements of the algorithm in different iteration periods. Then, in the local exploitation stage, the logarithmic spiral exploration mechanism is introduced to strengthen the local fine mining capacity around the optimal solution and further raise the convergence precision of the algorithm. Finally, an improved adaptive relative reflection strategy is added at the end of each iteration to enhance the population's capacity for escaping the local optimal and decrease the probability of local shackling. After theoretical analysis, it is proven that the improved algorithm proposed in this paper has the same time complexity as the basic marine predators algorithm. To analyze and verify the improved algorithm's performance, it is tested and compared with five representative comparison algorithms on the CEC2017 test suite in multiple dimensions, as well as with five challenging engineering design problems. The test results indicate that the improved algorithm proposed in this paper outperforms the other five representative algorithms in terms of optimization precision, convergence performance, and solution stability when solving complex multi-dimensional functions and engineering design optimization problems. Particularly, the superiority of its optimization performance is even more pronounced when dealing with high-dimensional complex functions.