This repository contains an implementation of a simple Liquid Neural Network (LNN). LNNs are dynamic neural networks inspired by biological neurons, which update their internal states over time using a combination of input stimuli and their previous state.
This implementation is designed for basic experimentation with the CIFAR-10 dataset, demonstrating how to process image data using a Liquid Neural Network.
- Custom Liquid Neural Network (LNN):
- Dynamic hidden state updates based on input and past states.
- Softmax output for classification tasks.
- Supports CIFAR-10 Dataset:
- Normalization and preprocessing included.
- Predictions on raw CIFAR-10 data.
- Python: Version 3.7 or higher
- Dependencies:
numpytensorflow