autodetect interpretation parameter using artificial intelligence

[gogo2464]

Is your feature request related to a problem?

I used to be so frustruated when I started urh because I had little knowledge and it was complicated to subdivide using apply bandpass filter in the interpretation paramter and then to change scale and zoom.

Describe the solution you'd like

I am sure this could be avoided using artifical intelligence so that when the user click in the button autodetect parameter, the real feature to autocomplete the previous steps could be autocompleted.

Describe alternatives you've considered

we could:

• replace the old button feature with a more accurate way using AI
• add a new button for the AI way under the old
• or simply do not implement it at all because nobody wants to implement it

Additional context

I might choose to be assigned to this issue. But not immediately. Maybe this month. I have no AI experience but I am a very experienced programmer.

[andynoack]

Hey @gogo2464,

we had the same idea before implementing the rule-based solution that is currently deployed. There were basically two reasons against a ML based solution years ago. The first reason is that rule-based AI is just faster and since URH is a GUI-application response times really matter. As long as the precision of rule-based and AI solutions is similar (and we assume that) there is no reason to deploy a more complex and slower system. The main reason that we do not expect a better precision of ML-based solutions is lack of training data. Artificial data that could be generated does not produce solutions for many real world applications due to different error/interference characteristics. Gathering enough (and suitable) real world data is a lot of organisational work.