https://github.com/samyak2/fourier-dnn

Tensorflow 2.0 implementation of fourier feature mapping networks.
https://github.com/samyak2/fourier-dnn

fourier-features tensorflow

Last synced: about 2 months ago
JSON representation

Tensorflow 2.0 implementation of fourier feature mapping networks.

• Host: GitHub
• URL: https://github.com/samyak2/fourier-dnn
• Owner: Samyak2
• License: apache-2.0
• Created: 2020-07-27T16:54:06.000Z (almost 6 years ago)
• Default Branch: master
• Last Pushed: 2020-12-10T09:50:02.000Z (over 5 years ago)
• Last Synced: 2025-02-27T02:51:05.349Z (over 1 year ago)
• Topics: fourier-features, tensorflow
• Language: Python
• Homepage:
• Size: 915 KB
• Stars: 1
• Watchers: 4
• Forks: 3
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# fourier-dnn

![Tests](https://github.com/Samyak2/fourier-dnn/workflows/Tests/badge.svg)

Tensorflow 2.0 implementation of fourier feature mapping networks.

Based on the paper ["Fourier Features Let Networks Learn High Frequency Functions in Low Dimensional

Domains"](https://arxiv.org/abs/2006.10739).

# Results

Results after training fourier feature mapping models for around 100-200 epochs.

![](./output_g_0_test.png)

![](./output_g_1_test.png)

![](./output_g_2_test.png)

# Setup

 - Git clone

 - Install the package locally (the `-e` flag means that any changes you make here in the code will be

    reflected in the package)

    ```bash

    pip install -e .

    ```

# Usage

Check out the [2D image regression example](fourier_dnn/2d_image_regression.py) for an example

usage.

Run the example using:

```bash

python ./fourier_dnn/2d_image_regression.py 3 100

```

where `3` is the image index from the dataset (of 16 images) and `100` is the number of epochs.

(Dataset taken from the [original

implementation](https://github.com/tancik/fourier-feature-networks))

Use `fourier_dnn.ffm_mlp.FourierMLP` to make a complete fourier feature mapping network.

For example:

```python

model = FourierMLP(10, 256, 3,

                   gaussian=True, staddev=16,

                   num_units_FFM=256)

```

Use `fourier_dnn.metrics.PSNR` as a metric for `model.compile` to get PSNR values on every epoch.

# Weights and Biases reports

 - [Effect of standard

deviation](https://app.wandb.ai/malthael/fourier-feature-dnn/reports/Effect-of-Standard-Deviation-on-Fourier-Feature-Networks--VmlldzoxODg1NDc)

 - [Effect of learning

   rate](https://app.wandb.ai/samyak/fourier-dnn/reports/Effect-of-learning-rate--VmlldzoxODg1Mjg)

 - [Effect of number of Dense

   units](https://app.wandb.ai/samyak/fourier-dnn/reports/Effect-of-number-of-Dense-units--VmlldzoxODg1NjQ)

 - [Effect of standard deviation

   2](https://app.wandb.ai/samyak/fourier-dnn/reports/Effect-of-number-of-Dense-units--VmlldzoxODg1NjQ)

 - [Effect of number of

   layers](https://app.wandb.ai/entherealenvy/fourier-dnn-fourier_dnn/reports/Fourier-DNN--VmlldzoxODg1MzI)