This new research from MIT and ETH Zurich is speeding up the solution of complex optimization problems using artificial intelligence (AI) and machine learning. By using a mixed-integer linear programming (MILP) solver, companies can find the best solution for routing problems more quickly and efficiently. This new data-driven approach can accelerate MILP solvers by 30 to 70 percent without sacrificing accuracy. The research has applications in various fields, including ride-hailing services, electric grid operations, and resource allocation problems. The team behind this innovation is also looking to apply this approach to even more complex optimization problems in the future.
Speeding Up Optimization Problems with AI
Companies like FedEx often use specialized software called a mixed-integer linear programming (MILP) solver to efficiently route holiday packages.
This software uses generic algorithms to find the best solution, but it can take hours or even days to arrive at a solution, leading to suboptimal results and wasted time.
Using Machine Learning to Solve Complex Problems
Researchers from MIT and ETH Zurich have used machine learning to speed up MILP solvers.
They identified a key intermediate step in MILP solvers that slows down the process and used a filtering technique to simplify this step. Then, they used machine learning to find the optimal solution for a specific type of problem. This approach can be tailored to a company’s own data to get the best solution for their specific problem.
The Future of Optimization with AI
This new data-driven approach has accelerated MILP solvers between 30 and 70 percent without any drop in accuracy. It has potential applications in various industries, and the researchers are looking to apply this approach to even more complex optimization problems in the future.
This research is supported by Mathworks, the National Science Foundation (NSF), the MIT Amazon Science Hub, and MIT’s Research Support Committee.