I guess these traces should depend on users network (location/latency/throughput/tor-enabled) and CPU. Did you use these as additional features for your model? Or was it a single configuration being tested?
In an earlier version of this project, I sent my code to 30 of my friends (all with different locations, networks, OS, etc) and had them collect a bunch of traces, and we found that the models lose a little accuracy when predicting traces on other machines, but not too much (maybe low 90s to low 80s, can be worse depending on the machines). We decided to focus more on the system interrupts/side channel analysis for this paper but we think that would be great follow-up work.