Well I believe the challenge here is you cannot have an alert on a process if it's not monitored in advance.
if you have the suspects in advance have those monitored by the OS KMs
if NOT, then you can achieve some of it using Scripting KM.
when having that installed, then one of the samples of that KM is called hogging-process.sh
that script is monitoring for the overall CPU% of the host, and it also annotates the parameter with the list of 10 highest consuming CPU processes.
if you run that Script for long and it reports High CPU of the host say > 75%, then you can look at the annotation (also existing in the event notes), and see the processes that contributed the most to that high number.
you can have that script run every 10 seconds or so, maybe once a minute and it will pick up the "CPU hogging" processes