View all files | ||||
The new IDK-based clustering algorithm, called IDKC, makes full use of the distributional kernel for trajectory similarity measuring and clustering. IDKC identifies non-linearly separable clusters with irregular shapes and varied densities in linear time.
All datasets are stored in ./datasets as .mat files, containing trajectory data and labels.
You can use IDK to generate vector embeddings of trajectories. Run ./IDK/traj_embedding.py under current directory:
The embedding data is stored in ./embeddings. You can also use MDS to visualize the embedding result:
After generating the embedding of trajectories, run ./TIDKC/IDKC_traj.mlx to do clustering.