https://github.com/erupturatis/neuralnetworkplayground
A website in which you can customize, train and visualize changes in your neural networks as they learn. You can also upload your own datasets and change any parameters
https://github.com/erupturatis/neuralnetworkplayground
backpropagation d3 d3js d3js-graph machine-learning neural-network neural-network-visualization react reactjs rendering visualization
Last synced: 9 months ago
JSON representation
A website in which you can customize, train and visualize changes in your neural networks as they learn. You can also upload your own datasets and change any parameters
- Host: GitHub
- URL: https://github.com/erupturatis/neuralnetworkplayground
- Owner: erupturatis
- License: mit
- Created: 2022-11-21T16:08:05.000Z (over 3 years ago)
- Default Branch: master
- Last Pushed: 2023-07-21T19:41:36.000Z (almost 3 years ago)
- Last Synced: 2023-07-21T22:21:14.772Z (almost 3 years ago)
- Topics: backpropagation, d3, d3js, d3js-graph, machine-learning, neural-network, neural-network-visualization, react, reactjs, rendering, visualization
- Language: JavaScript
- Homepage: https://seal-app-4qpjq.ondigitalocean.app/Home
- Size: 381 KB
- Stars: 17
- Watchers: 3
- Forks: 1
- Open Issues: 4
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
🚀 NeuralNetworkPlayground
A better way of visualzing neural networks
# Overview
As the title implies, this is a website made in order to help visualize neural network in the process of training and backpropagating. You can currently:
- #### Create any neural network with any arhitecture
- #### Upload your own custom datasets
- #### Visualize the training and being able to replay previous epochs
- #### Save your model + training and loading it later
- #### Optimze network parameters and hyperparameters
- #### Change the cosmetics of how the network is displayed
## Looks
### Note !
#### This website was meant as a pet project so it might still have a lot of bugs. If you like this project and want to make it better feel free to open an issue or submit a PR.
## Roadmap
- Adding more activation function
- Adding Regularization
- Only doing forward propagation for the current network
- Visualizing the output neurons relation to input data on a 2d graph similar to [here](https://playground.tensorflow.org/#activation=tanh&batchSize=10&dataset=circle®Dataset=reg-plane&learningRate=0.03®ularizationRate=0&noise=0&networkShape=4,2&seed=0.71560&showTestData=false&discretize=false&percTrainData=50&x=true&y=true&xTimesY=false&xSquared=false&ySquared=false&cosX=false&sinX=false&cosY=false&sinY=false&collectStats=false&problem=classification&initZero=false&hideText=false)
- Being able to modify individual weights and neurons and deactivate
- Extending the website to [CNN](https://en.wikipedia.org/wiki/Convolutional_neural_network) or [RNN](https://en.wikipedia.org/wiki/Recurrent_neural_network) arhitectures
- A lot more
## Contributing
### Tech Stack
**Client:** React, Redux, TailwindCSS, Vite
**Server:** Node, Express
#### To start just clone the repo
- If you want to contribute to the frontend just run `npm install` in the client folder and then`npm run dev`. You will only need a `VITE_API_URL` variable in your .env file if you want to work with the server too.
- If you want to contribute to the backend run `npm install` in the server folder and then`npm run dev`. You will need the env file with the following
```
GITHUB_CLIENT_ID = ''
GITHUB_CLIENT_SECRET = ''
PORT = ''
SECRET =''
MONGO_URI = "mongodb+srv://username:@cluster0.wudiukh.mongodb.net/?retryWrites=true&w=majority"
MONGO_PASSWORD = ''
GOOGLE_CLIENT_ID = ''
GOOGLE_CLIENT_SECRET = ''
CLIENT_URL = ''
SERVER_URL = ''
```
and set up the oauth services properly. If you have any issues running the project just open a new issue and I will help you as fast as I can
### Apart from this, I hope you like the website!