We proposed to feed a neural network time-domain data generated from the FDTD model along with its ground-truth area function. Hopefully, the neural network will be able to learn the mapping scheme and generate the corresponding geometry from unknown time-domain data
.
CPSC554X_project/
├── 1segtube
│ ├── dataset
│ │ ├── acoustic_data.txt
│ │ ├── audio_1.mat
│ │ └── geometry_data.txt
│ ├── generate_dataset.py
│ └── mlp.py
├── #segtube
│ ├── dataset
│ │ ├── acoustic_data.txt
│ │ └── geometry_data.txt
│ ├── generate_dataset.py
│ └── mlp.py
└── README.md
- 1. Generate dataset
- 2. Train a neural network
- 3. Test the neural network
- 4. Generate a 2D geometry from unknown time-domain data
- 5. Write a report