libphysneurallib 0.1.0

Deep learning library for biosignal processing.

NeuralLib

NeuralLib is a Python library designed for advanced biosignal processing using neural networks. The primary objective is to establish a modular, efficient, generalizable framework for biosignal processing using DL. The core concept of NeuralLib revolves around creating, training, and managing neural network models and leveraging their components for transfer learning (TL). This allows for the reusability of pre-trained models or parts of them to create new models and adapt them to different tasks or datasets efficiently.

The library supports:

• Training and testing Architectures from scratch for specific biosignals processing tasks.
• Using trained models (ProductionModels) to process biosignals.
• Adding tested models to hugging face repositories to create new ProductionModels and share them with the community for public usage.
• Extracting trained components from production models using TLFactory.
• Combining, freezing, or further fine-tuning pre-trained components to trainTLModels.

Tutorials

Explore the tutorials/ folder for several hands-on examples demonstrating how to use the core functionalities of NeuralLib.

📖 Documentation

Comprehensive documentation is available here:
NeuralLib Documentation

Pre-trained Models

Collection of pre-trained models on Hugging Face:
NeuralLib DL Models for Biosignals