The values are tracking very closely, however the 'CPU Sampling by Process' appears to be less fine grained in it's data. I can't find anything in the documentation to differentiate these two.
CPU Usage seems to include Context Switches and allows you to filter by Thread Priority, thus it seems to be more related to threads while CPU Usage is more related to processes due to the aggregation that occurs.
See the list of Sample Analyzer Graphs at the left side,
when comparing take attention of the words "aggregated" and "context switches" is the description.