Artificial intelligence (AI) and machine learning (ML) are becoming more and more widely used, but up to this point, haven’t really made any inroads into ham radio. One of the reasons for this is that it’s hard. Another is that I’m not sure that we have enough of the data that AI/ML systems use to do what they do.
Even so, it’s coming. Michelle, W5NYV, has some ideas on the topic. Rather than try to summarize them, I’ll let her presentation to the latest QSO Today Virtual Ham Expo speak for itself.
Filtering with neural networks
On a related topic, Kurt, AD0WE, of Morse Code Ninja fame, posted an interesting message to the Long Island CW Club mailing list. Titled “Bleeding edge tech — Filtering with Neural Networks,” it asks the question, “Can the noise of received HF transmissions be removed using Artificial Intelligence?”
“Yes!” he answers, “The amateur radio community could create a version of DeepFilterNet that is user-friendly for amateur radio operators and has all the knobs to tweak its performance in real-time. It requires modest hardware for real-time filtering.”
This looks like it could be an interesting project for someone with the time and talent to pursue it.
In the video, Michelle points out that whether we like it or not, artificial intelligence is coming. I agree with her, and the sooner, someone, be it one of the ham radio manufacturers or an individual ham, figures out how to take advantage of it, the better.