A Mac software that visualizes BP Neural Network training progress.
- Accept files as data input and trains a neural network to fit the data
- Provide user interface to customize neural network's architecture, learning rate, lambda, and number of iteration
- Show change of cost function, training set precision and test set precision throughout the training process
- Support parameter update (learning rate, lambda, and number of iteration) during training
- Cost Function Tab: Shows change of cost function with x-axis as number of iteration and y-axis as value (with lowest side 0 and highest side cost function before the first iteration)
- Precision Tab: Shows change of precision throughout training process. Red line denotes training set precision; blue line denotes test set precision
- Test Set Tab: Visualizes which data in the whole dataset is identified correctly. White box indicates correct and black means incorrect
- Table: Presents cost function and precision in raw data format
- Network: Provides UI to change the architecture of the neural network
- Data: Provides UI to browse data from files and specify file format
- Training: Provides UI to customize parameters related to training process
- Start and Stop Buttons: Allows user to start or terminate the training
- Size of data and dataset cannot be automatically read; incorrectly inputing values might cause program error
- Does not support training for regression problems yet
- Any other errors that are yet to be found
This project is for self-learning purposes only and is not guaranteed correct in all cases. The author of this project is not responsible for any negative outcome that using this project gives rise to.