GMDNet
GMDNet: A Graph-Based Mixture Density Network for Estimating Packages’ Multimodal Travel Time Distribution
Usage
Use the following command to run the code:
cd algorithm/GMDNet
python train.py
Repo Structure
The structure of our code and description of important files are given as follows:
│────algorithm/
│ ├────gmdnet/: code of GMDNet.
│────data/: a subset of data used for train, validation, and test.
│────my_utils/
│ ├────util.py
Parameters:
| Name | Type | Description |
|---|---|---|
| hidden_dim | int | number of hidden units. |
| n_gaussians | int | number of Gaussian components. |
| att_hidden_size | int | number of hidden units in the self-attention. |
| num_of_attention_heads | int | num_of_attention_heads. |
| num_layers | int | number of GNN layers. |
| dirichlet_alpha | int | setting of Dirichlet regularizer. |
| batch_size | int | number of samples per batch. |
| lr | float | learning rate for training. |
| early_stop | int | stop training when a monitored metric has stopped improving. |
| num_epochs | int | number of passes over the training data. |