Visualizing Feasible Regions for Optimization Problems on High-Dimensional Permutations using Dimensionality Reduction Methods
Igor Grebennik, О. С. Чорна, Inna Urniaieva · 2023
In this paper, we undertake an exploration into the application of contemporary dimensionality reduction techniques to address traditional combinatorial optimization problems. Our research introduces the utilization of the t-Distributed Stochastic Neighbor Embedding (t-SNE) method to visualize the admissible solution regions within high-dimensional permutations. The main objective is to mitigate the challenges posed by combinatorial explosion. Through our investigation, we discover that the proposed approach yields promising outcomes, offering significant insights and enhancing the comprehension of the solution space for high-dimensional permutations. This research contributes to the advancement of understanding and analysis within the field of combinatorial optimization, highlighting the potential benefits of employing modern dimensionality reduction methods in this context.