Home > Datasets, experiments, what i've been doing > Results for NGMAST11/NEMO

Results for NGMAST11/NEMO

We have decided not to submit to NGMAST11 conference as the results we have were not really ready, also I found an error in the config setup for KCLIQUE BubbleH. I have therefore run all of the simulations from scratch, just to make sure it is all correct. In this re-run, I ran the HGCE algorithms far too many times for each dataset, so there is a very large number of results for the HGCE algorihtm (~700 for each dataset). We plan to submit to the Finding NEMO workshop instead.

To visualise the results so far, we have plotted the best  (see note below) results for BubbleRAP and BubbleH:

matrix

BUBBLERap and BubbleH compared for Delivery Ratio, Cost and Latency

To determine the ‘best’ run for algorithms with multiple parameters, we first filtered the results to the best 10 for latency, then ordered by delivery ratio followed by cost, this should mean that a good delivery ratio is still shown for good delivery latency, but avoids results where a abnormally low delivery ratio is chosen with a very poor delivery ratio.

  1. No comments yet.
  1. No trackbacks yet.