sigqnn 0.0.1.dev0

Quantized Neural Networks for Signal Processing (PLACEHOLDER - NOT FUNCTIONAL)

SigQNN - Signal Processing with Quantized Neural Networks

** DEVELOPMENT IN PROGRESS - NOT YET FUNCTIONAL **

This PyPI package name is reserved for an upcoming open-source toolkit for:

• Automatic Modulation Classification (AMC) with quantized CNNs.
• FPGA deployment via FINN/QONNX .
• RadioML dataset integration.

Current Status: Pre-alpha placeholder. Do not install.

Expected Release: Q3 2026

Repository: https://github.com/rothej/sigqnn

---

About the Project

SigQNN is being developed from research on efficient neural networks for wireless signal classification. The toolkit will provide:

• Pre-configured CNN architectures (VGG-like, more?).
• Quantization-aware training (Brevitas integration).
• Structured pruning for FPGA synthesis.
• RadioML 2016.10a dataset loader with Zenodo mirror.
• ONNX/QONNX export for FINN compiler.

Based on:

• Rothe, J. (2024). Quantization and Pruning of Convolutional Neural Networks for Efficient FPGA Implementation of Digital Modulation Detection Firmware [Master's Thesis]
• Dataset: O'Shea & West (2016), RadioML 2016.10a (CC BY-NC-SA 4.0)

---

This placeholder ensures the package name is reserved during active development.