TorchSig, a toolkit for applying deep-learning applications to wireless signals, was released last year and presented at GRCon. The toolkit has since undergone numerous changes to improve reliability and usability. This includes new examples for using deep-learning for tasks in spectrum awareness, new utilities for curating datasets, speedups in synthetic dataset generation and speedups in data augmentations.
|Talk Length||15 Minutes|
|Link to Open Source Code||https://github.com/TorchDSP/torchsig|