Machine learning method for combinatorial optimization problems

Optimization is a very classical research field, and many problems in engineering, management, and other fields are essentially optimization problems. At present, the rapid development of machine learning methods represented by deep learning and reinforcement learning has brought key promotion and development in many subject areas. Optimization is one of the most important supporting disciplines of machine learning, and machine learning is ultimately transformed into solving an optimization problem, which belongs to the research content of science for AI (artificial intelligence). In turn, AI technologies, including machine learning, are already having a significant impact on the development of science and will form a new paradigm of scientific research—AI for science. The artificial intelligence method of optimization problems is to design a fast and efficient algorithm of optimization problems using the theory of artificial intelligence. In this paper, we summarize the research progress of machine learning methods for combinatorial optimization in recent years, and deeply analyze the difficulties of solving combinatorial optimization problems and the bottlenecks of artificial intelligence solving methods. Some important research directions are given based on the current research status of image processing, game theory, reinforcement learning, and other popular fields.