Memristive Boltzmann machine: A hardware accelerator for combinatorial optimization and deep learning

Combinatorial optimization is a branch of discrete mathematics that is concerned with finding the optimum element of a finite or countably infinite set. An enormous number of critical problems in science and engineering can be cast within the combinatorial optimization framework, including classical problems such as the traveling salesman, integer linear programming, knapsack, bin packing, and scheduling, as well as numerous optimization problems in machine learning and data mining. Because many of these problems are NP-hard, heuristic algorithms commonly are used to find approximate solutions for even moderately sized problem instances.