VortexNet: Neural network based on fluid dynamics

This repository contains toy implementations of the concepts introduced in the research paper VortexNet: Neural Computing through Fluid Dynamics. These examples demonstrate how PDE-based vortex layers and fluid-inspired mechanisms can be integrated into neural architectures, such as autoencoders for different datasets.

Note: These are toy prototypes for educational purposes and are not intended as fully optimized or physically precise fluid solvers.

• vortexnet_mnist.py:
  A demonstration script for building and training a VortexNet Autoencoder on the MNIST dataset.
• vortexnext_image.py:
  An advanced script for building and training a VortexNet Autoencoder on custom image datasets with enhanced features like data augmentation and latent space interpolation.

Ensure you have Python 3.8+ installed. Install the required Python packages using pip:

• MNIST Dataset:
  The MNIST dataset will be automatically downloaded by vortexnet_mnist.py if not already present.

• Custom Image Dataset:
  For vortexnext_image.py, place your images (JPEG, PNG, or JPEG formats) inside the my_data/ directory.

This script builds and trains a VortexNet Autoencoder on the MNIST dataset.

Usage:

This advanced script builds and trains a VortexNet Autoencoder on custom image datasets with enhanced features.

Usage:

• Configuration Files:
  Ensure the configuration file (config_image.yaml) is properly set up before running the scripts.

• Output Directory:
  All outputs, including logs, reconstructed images, and model checkpoints, are saved in the output_dir specified in the respective configuration files.

• TensorBoard:
  For monitoring training progress, you can launch TensorBoard pointing to the output_dir