Shortest Path Problem

该数据集是一个为5x5网格中的最短路径问题生成的合成数据集，目标是从西南角导航至东北角。数据集中的特征向量是从多元高斯分布中抽取的。此外，该数据集允许模型规范度的不同级别，分别为1、2、4、6和8，同时噪声是从区间[0.5, 1.5]的均匀分布中抽取的。对于不同级别的模型规范度，分别提供了1000、250和10000个训练、验证和测试实例。该数据集的任务是最短路径优化。

This dataset is a synthetic dataset generated for the shortest path problem in a 5x5 grid, with the goal of navigating from the southwest corner to the northeast corner. The feature vectors in the dataset are sampled from a multivariate Gaussian distribution. Additionally, this dataset supports multiple levels of model regularization, specifically 1, 2, 4, 6 and 8, while the noise is sampled from a uniform distribution over the interval [0.5, 1.5]. For each level of model regularization, 1000 training, 250 validation and 10000 test instances are respectively provided. The task of this dataset is shortest path optimization.